# US 9,002,759 B2 · Data mining technique with maintenance of fitness history

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 9,002,759 B2
- Document ID: US9002759B2
- Status: Granted
- Kind: B2
- Application number: 13/358,381
- Filed: 2012-01-25
- Published: 2013-05-16
- Granted: 2015-04-07
- Earliest claimed: 2011-11-14
- Assignees: Sentient Technologies (Barbados) Limited
- Applicants: Babak Hodjat; Hormoz Shahrzad
- Inventors: Babak Hodjat; Hormoz Shahrzad
- Web page: https://patentgrep.com/patent/US9002759B2

## CPC

- G06N 3/126: Evolutionary algorithms, e.g. genetic algorithms or genetic programming
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

Roughly described, a data mining system includes a candidate gene database in which each candidate individual has a respective fitness estimate. A gene pool processor tests individuals from the candidate gene pool on training data and updates the fitness estimate of the individuals. A gene harvesting module deploys selected individuals from the gene pool. The gene pool processor includes a competition module which selects individuals for discarding in dependence upon their updated fitness estimate. The system maintains a fitness training history for each of the candidate individuals, identifying the data samples on which the individual has been tested. The historical information can be used to assist in any one or more of the following: competition among the individuals, avoiding re-testing of an individual on the same data sample, removing duplicate test data before merging fitness evaluations, improving gene pool diversity, and selecting individuals for deployment.

## Claims

### Claim 1

A data mining system, for use with a data mining training database containing a plurality of data samples, comprising:
a computer system having a memory having a candidate gene database identifying a pool of candidate individuals, each of the candidate individuals identifying a plurality of conditions and at least one corresponding proposed output in dependence upon the conditions,
a gene pool processor which:
tests each individual in a testing subset of at least one of the candidate individuals, each of the tests applying the conditions of the respective individual to a respective subset of the data samples in the training database to propose a result, each individual in the testing subset being tested on at least one data sample and at least one of the individuals in the testing subset being tested on more than one data sample;
stores, in association with each of the candidate individuals in the testing subset, a respective first history identifying the data samples on which the respective individual has been tested; and
calculates an overall fitness estimate for each of the individuals in the testing subset, in dependence upon the results proposed by the respective individual when the conditions of the respective individual were applied to each of the data samples indicated in the first history associated with the respective individual,
the gene pool processor further including a competition module which selects individuals for discarding in dependence upon comparisons among their respective overall fitness estimates,
the computer system further having a gene harvesting module providing for deployment selected ones of the remaining individuals from the pool of candidate individuals.

### Claim 2 (depends on 1)

The system according to claim 1, wherein the testing performed by the gene pool processor of each individual in a testing subset comprises testing in a manner which precludes testing each given individual in the testing subset from applying the conditions of the respective individual to any data sample already identified in the first history associated with the given individual.

### Claim 3 (depends on 1)

The system according to claim 1, wherein the testing performed by the gene pool processor of each individual in a testing subset includes the computer system:
identifying, in dependence upon the first history associated with a given individual, a given data sample not already identified in the first history associated with the given individual;
retrieving the given data sample from the data mining training database; and
testing the given individual on the given data sample.

### Claim 4 (depends on 1)

The system according to claim 1, wherein the testing performed by the gene pool processor of each individual in a testing subset comprises the computer system:
receiving a block of at least one data sample;
determining in dependence upon the first history associated with a given individual whether each data sample in the block is already identified in the first history associated with the given individual; and
testing the given individual on only those data samples in the block, if any, which are not already identified in the first history associated with the given individual.

### Claim 5 (depends on 1)

The system according to claim 1, wherein the gene pool processor, after storing a first history and before calculating an overall fitness estimate:
further tests a particular individual in the testing subset on at least one further data sample; and
stores, in association with the particular individual a further history identifying each of the at least one further data sample,
and wherein in calculating an overall fitness estimate, the gene pool processor:
determines a union of the data samples identified in the first history and in the further history, the union excluding all duplicate data samples;
forms, in dependence upon the union, a combined history which identifies the data samples on which the particular individual has been tested; and
calculates the overall fitness estimate for the particular individual in dependence upon the results proposed by the particular individual when the conditions of the particular individual were applied to each of the data samples indicated in the combined history associated with the particular individual.

### Claim 6 (depends on 5)

The system according to claim 5, wherein in determining a union and forming a combined history, collectively, the computer system, for each particular data sample identified in the further history:
determines whether the particular data sample is already identified in the first history associated with the particular individual; and
adds an identification of the particular data sample to a history associated with the particular individual only if the particular data sample is not already identified in the first history associated with the particular individual.

### Claim 7 (depends on 5)

The system according to claim 5, wherein in storing a first history, the gene pool processor stores, in association with each of the candidate individuals in the testing subset, the respective first history so as to identify both the data samples on which the respective individual has been tested and the performance of the respective individual on each of such data samples,
wherein in forming a combined history, the gene pool processor forms, in dependence upon the union, the combined history so as to identify both the data samples on which the particular individual has been tested and the performance of the respective individual on each of such data samples,
and wherein in calculating the overall fitness estimate for the particular individual the gene pool processor calculates the overall fitness estimate for the particular individual in dependence upon the performance indications in the combined fitness history.

### Claim 8 (depends on 1)

The system according to claim 1, wherein in selecting individuals for discarding in dependence upon comparisons among their respective overall fitness estimates, the gene pool processor selects the individuals for discarding in further dependence upon a characteristic of the individuals' first histories other than their respective overall fitness estimates.

### Claim 9 (depends on 1)

The system according to claim 1, wherein in storing a first history, the gene pool processor stores, in association with each of the candidate individuals in the testing subset, the respective first history so as to identify both the data samples on which the respective individual has been tested and the performance of the respective individual on each of such data samples.

### Claim 10 (depends on 9)

The system according to claim 9, wherein in the selection of individuals for discarding in dependence upon comparisons among their respective overall fitness estimates, the competition module selects the individuals for discarding in further dependence upon a characteristic of the individuals' first histories other than their respective overall fitness estimates.

### Claim 11 (depends on 9)

The system according to claim 9, wherein in calculating an overall fitness estimate for each of the individuals in the testing subset, the gene pool processor calculates the overall fitness estimate for each particular one of the individuals in the testing subset in further dependence upon a characteristic of the first history associated with the particular individual other than an unweighted average performance of the particular individual.

### Claim 12 (depends on 9)

The system according to claim 9, wherein in calculating an overall fitness estimate for each of the individuals in the testing subset the gene pool processor:
calculates a preliminary fitness estimate for each particular one of the individuals in the testing subset; and
weights the preliminary fitness estimate for at least some of the individuals in the testing subset in dependence upon an additional characteristic of the first history associated with the particular individual.

### Claim 13 (depends on 12)

The system according to claim 12, wherein in weighting the preliminary fitness estimate for at least some of the individuals in the testing subset, the gene pool processor weights the preliminary fitness estimate for only those of the individuals in the testing subset which have been tested on at least a predetermined number N>1 of data samples.

### Claim 14 (depends on 12)

The system according to claim 12, wherein the memory further includes data identifying layer parameters for a number T>0 of gene pool experience layers L1-LT in an elitist pool, the layer parameters for each i'th one of the layers L1-LT identifying a respective minimum testing experience level ExpMin(Li) for the layer, ExpMin(L1)>1 and, if T>1, ExpMin(Li)>ExpMin(Li-1) for each i>1,
and wherein in weighting the preliminary fitness estimate for at least some of the individuals in the testing subset, the gene pool processor weights the preliminary fitness estimate for only those of the individuals in the testing subset which have been tested on at least a predetermined number N of data samples, where N is equal to one of the ExpMin(Li).

### Claim 15 (depends on 12)

The system according to claim 12, wherein in selecting individuals for discarding in dependence upon comparisons among their respective overall fitness estimates, the competition module selects individuals for discarding in dependence upon comparisons among their respective weighted fitness estimates.

### Claim 16 (depends on 12)

The system according to claim 12, wherein in providing for deployment of selected ones of the remaining individuals from the pool of candidate individuals, the gene harvesting module provides for deployment individuals from the pool of candidate individuals selected in dependence upon comparisons among their respective weighted fitness estimates.

### Claim 17 (depends on 9)

The system according to claim 9, wherein in selecting individuals for discarding in dependence upon comparisons among their respective overall fitness estimates, the gene pool processor selects the individuals for discarding in further dependence upon a comparison of a particular characteristic of one individual's first history other than its overall fitness estimate, with the particular characteristic of a different individual's first history.

### Claim 18 (depends on 9)

The system according to claim 9, wherein in providing for deployment of selected ones of the remaining individuals from the pool of candidate individuals, the gene harvesting module selects for deployment individuals from the pool of candidate individuals in dependence upon an additional characteristic of the first histories associated with the individuals in the pool of candidate individuals.

### Claim 19 (depends on 1)

The system according to claim 1, wherein each of the data samples in the training database indicates historical market data at a plurality of times in a trading day, including the price of a security at the plurality of times,
wherein each individual indicates a plurality of conditions resolvable in dependence upon the data in a given one of the samples, the outputs identified by the individuals including security trading instructions,
wherein in testing each individual in a testing subset, for each trial of a given one of the individuals being tested, the gene pool processor hypothetically performs all the trading instructions made by the given individual on a respective given one of the samples throughout the trading day of the given sample,
and wherein in calculating an overall fitness estimate, the gene pool processor updates the fitness estimate associated with the given individual in dependence upon the given individual's profit or loss for the trading day of the given sample.

### Claim 20 (depends on 1)

The system according to claim 1, wherein the computer system comprises a server and a collection of at least one client device, and wherein in testing each individual in a testing subset of at least one of the candidate individuals:
the server delegates the testing to the client devices, including the server
delegating testing of a first one of the individuals in the testing subset to a first one of the client devices and
receiving back the first individual, in association with a first client-centric history identifying the data samples on which the first individual has been tested and the performance of the first individual on each of such data samples, the first client-centric history including at least all the data samples on which the first individual has been tested by the first client device;
and wherein in storing, in association with each of the candidate individuals in the testing subset, the gene pool processor updates a server-centric history associated with the first individual to identify the data samples and the performance indications from the first client-centric history received back from the first client device.

### Claim 21 (depends on 20)

The system according to claim 20, wherein the server, after delegating testing of the first individual to the first client device and before updating a server-centric history associated with the first individual:
delegates to a second one of the client devices additional testing of the first individual; and
receives back the first individual from the second server, in association with a second client-centric history identifying a second subset of at least one of the data samples in the training database on which the first individual has been tested by the second client device and the performance of the first individual on each of such data samples in the second subset, the second client-centric history including at least all the data samples on which the first individual has been tested by the second client device,
wherein in storing, in association with each of the candidate individuals in the testing subset, the server updates the server-centric history associated with the first individual to identify the data samples and the performance indications from both the first and second client-centric histories,
and in updating the server-centric history, the server:
determines a union of the data samples identified in the first and second histories, the union excluding all duplicate data samples; and
forms the first history for the first individual in dependence upon the union.

### Claim 22

A client computer system for a data mining system, for use with a data mining training database containing a plurality of data samples, comprising:
a processing subsystem;
a memory having a candidate gene database identifying a client-centric pool of candidate individuals, each of the candidate individuals identifying a plurality of conditions and at least one corresponding proposed output in dependence upon the conditions,
a client gene pool processor which:
tests each individual in a testing subset of at least one of the candidate individuals, each of the tests applying the conditions of the respective individual to a respective subset of the data samples in the training database to propose a result, each individual in the testing subset being tested on at least one data sample, the testing being performed in a manner that precludes testing each given individual in the testing subset from applying the conditions of the respective individual to any data sample already identified in the first history associated with the given individual;
stores, in association with each of the candidate individuals in the testing subset, a respective first history identifying the data samples on which the respective individual has been tested, at least so far as is known by the computer system; and
calculates an client-centric overall fitness estimate for each of the individuals in the testing subset, in dependence upon the results proposed by the respective individual when the conditions of the respective individual were applied to each of the data samples indicated in the first history associated with the respective individual,
the client computer system further including a competition module which selects individuals for discarding in dependence upon comparisons among their respective client-centric overall fitness estimates; and
the client computer system further including a gene harvesting module which forwards to a central server infrastructure for potential deployment or further testing, selected ones of the remaining individuals from the client-centric pool of candidate individuals.

### Claim 23 (depends on 22)

The system according to claim 22, wherein in testing each individual in a testing subset, the client gene pool processor:
identifies, in dependence upon the first history associated with a given individual, a given data sample not already identified in the first history associated with the given individual;
retrieves the given data sample from the data mining training database; and
tests the given individual on the given data sample.

### Claim 24 (depends on 22)

The system according to claim 22, wherein in testing each individual in a testing subset, the client gene pool processor:
receives a block of at least one data sample;
determines in dependence upon the first history associated with a given individual whether each data sample in the block is already identified in the first history associated with the given individual; and
tests the given individual on only those data samples in the block, if any, which are not already identified in the first history associated with the given individual.

### Claim 25 (depends on 22)

The system according to claim 22, wherein in storing, the client gene pool processor stores, in association with each of the candidate individuals in the testing subset, the respective first history so as to identify both the data samples on which the respective individual has been tested and the performance of the respective individual on each of such data samples.

### Claim 26 (depends on 25)

The system according to claim 25, wherein in selecting individuals for discarding in dependence upon comparisons among their respective overall fitness estimates, the competition module selects the individuals for discarding in further dependence upon a characteristic of the individuals' first histories other than their respective overall fitness estimates.

### Claim 27

A computer-implemented data mining method, for use with a data mining training database containing a plurality of data samples, comprising the steps of:
providing a computer system having a memory having a candidate gene database identifying a pool of candidate individuals, each of the candidate individuals identifying a plurality of conditions and at least one corresponding proposed output in dependence upon the conditions,
the computer system testing each individual in a testing subset of at least one of the candidate individuals, each of the tests applying the conditions of the respective individual to a respective subset of the data samples in the training database to propose a result, each individual in the testing subset being tested on at least one data sample and at least one of the individuals in the testing subset being tested on more than one data sample;
storing, in association with each of the candidate individuals in the testing subset, a respective first history identifying the data samples on which the respective individual has been tested;
calculating an overall fitness estimate for each of the individuals in the testing subset, in dependence upon the results proposed by the respective individual when the conditions of the respective individual were applied to each of the data samples indicated in the first history associated with the respective individual;
the computer system selecting individuals for discarding in dependence upon comparisons among their respective overall fitness estimates; and
providing for deployment of selected ones of the remaining individuals from the pool of candidate individuals.

### Claim 28 (depends on 27)

The method according to claim 27, wherein the step of the computer system testing each individual in a testing subset comprises performing a step which precludes testing each given individual in the testing subset from applying the conditions of the respective individual to any data sample already identified in the first history associated with the given individual.

### Claim 29 (depends on 27)

The method according to claim 27, further comprising the steps, after the step of storing a first history and before the step of calculating an overall fitness estimate, of:
further testing a particular individual in the testing subset on at least one further data sample; and
storing, in association with the particular individual a further history identifying each of the at least one further data sample,
and wherein the step of calculating an overall fitness estimate comprises the steps of:
determining a union of the data samples identified in the first history and in the further history, the union excluding all duplicate data samples;
forming, in dependence upon the union, a combined history which identifies the data samples on which the particular individual has been tested; and
calculating the overall fitness estimate for the particular individual in dependence upon the results proposed by the particular individual when the conditions of the particular individual were applied to each of the data samples indicated in the combined history associated with the particular individual.

### Claim 30 (depends on 29)

The method according to claim 29, wherein the step of storing comprises the step of storing, in association with each of the candidate individuals in the testing subset, the respective first history so as to identify both the data samples on which the respective individual has been tested and the performance of the respective individual on each of such data samples,
wherein the step of forming a combined history comprises the step of forming, in dependence upon the union, the combined history so as to identify both the data samples on which the particular individual has been tested and the performance of the respective individual on each of such data samples,
and wherein the step of calculating the overall fitness estimate for the particular individual comprises calculating the overall fitness estimate for the particular individual in dependence upon the performance indications in the combined fitness history.

### Claim 31 (depends on 27)

The method according to claim 27, wherein the step of the computer system selecting individuals for discarding in dependence upon comparisons among their respective overall fitness estimates, selects the individuals for discarding in further dependence upon a characteristic of the individuals' first histories other than their respective overall fitness estimates.

### Claim 32 (depends on 27)

The method according to claim 27, wherein the step of storing comprises the step of storing, in association with each of the candidate individuals in the testing subset, the respective first history so as to identify both the data samples on which the respective individual has been tested and the performance of the respective individual on each of such data samples.

### Claim 33 (depends on 32)

The method according to claim 32, wherein the step of calculating an overall fitness estimate for each of the individuals in the testing subset comprises the steps of:
calculating a preliminary fitness estimate for each particular one of the individuals in the testing subset; and
weighting the preliminary fitness estimate for only those of the individuals in the testing subset which have been tested on at least a predetermined number N>1 of data samples, in dependence upon an additional characteristic of the first history associated with the particular individual,
and wherein the step of selecting individuals for discarding in dependence upon comparisons among their respective overall fitness estimates comprises selecting individuals for discarding in dependence upon comparisons among their respective weighted fitness estimates.

### Claim 34 (depends on 32)

The method according to claim 32, wherein the step of calculating an overall fitness estimate for each of the individuals in the testing subset comprises the steps of:
calculating a preliminary fitness estimate for each particular one of the individuals in the testing subset; and
weighting the preliminary fitness estimate for at least some of the individuals in the testing subset in dependence upon an additional characteristic of the first history associated with the particular individual,
and wherein the step of providing for deployment of selected ones of the remaining individuals from the pool of candidate individuals, comprises the step of providing for deployment individuals from the pool of candidate individuals selected in dependence upon comparisons among their respective weighted fitness estimates.

### Claim 35 (depends on 27)

The method according to claim 27, wherein the computer system comprises a server and a collection of at least one client device, and wherein the step of testing comprises the steps of:
the server delegating the testing to the client devices, including delegating testing of a first one of the individuals in the testing subset to a first one of the client devices; and
after the first client device has tested the first individual on at least one data sample, the server receiving back the first individual, in association with a first client-centric history identifying the data samples on which the first individual has been tested and the performance of the first individual on each of such data samples, the first client-centric history including at least all the data samples on which the first individual has been tested by the first client device,
and wherein the step of storing, in association with each of the candidate individuals in the testing subset, a respective first history identifying the data samples on which the respective individual has been tested, comprises the step of updating a server-centric history associated with the first individual to identify the data samples and the performance indications from the first client-centric history received back from the first client device.

### Claim 36 (depends on 35)

The method according to claim 35, further comprising the steps of:
the server, after delegating testing of the first individual to the first client device and before updating a server-centric history associated with the first individual, delegating to a second one of the client devices additional testing of the first individual, the second client device testing the first individual on a second subset of at least one of the data samples in the training database to propose results; and
the server receiving back the first individual from the second server, in association with a second client-centric history identifying the data samples on which the first individual has been tested by the second client device and the performance of the first individual on each of such data samples, the second client-centric history including at least all the data samples on which the first individual has been tested by the second client device,
wherein the step of storing, in association with each of the candidate individuals in the testing subset comprises the step of updating the server-centric history associated with the first individual to identify the data samples and the performance indications from both the first and second client-centric histories,
and wherein the step of updating the server-centric history comprises the steps of:
determining a union of the data samples identified in the first and second histories, the union excluding all duplicate data samples; and
forming the first history for the first individual in dependence upon the union.

### Claim 37

A computer-implemented data mining method, for use with a data mining training database containing a plurality of data samples, comprising the steps of:
providing a computer system having a memory having a candidate gene database identifying a pool of candidate individuals, each of the candidate individuals identifying a plurality of conditions and at least one corresponding proposed output in dependence upon the conditions,
the computer system testing each individual in a testing subset of at least one of the candidate individuals, each of the tests applying the conditions of the respective individual to a respective subset of the data samples in the training database to propose a result, each individual in the testing subset being tested on at least one data sample, the testing being performed in a manner that precludes testing each given individual in the testing subset from applying the conditions of the respective individual to any data sample already identified in the first history associated with the given individual;
storing, in association with each of the candidate individuals in the testing subset, a respective first history identifying the data samples on which the respective individual has been tested, at least so far as is known by the computer system;
calculating an overall computer system-centric fitness estimate for each of the individuals in the testing subset, in dependence upon the results proposed by the respective individual when the conditions of the respective individual were applied to each of the data samples indicated in the first history associated with the respective individual;
the computer system selecting individuals for discarding in dependence upon comparisons among their respective overall fitness estimates; and
forwarding to a central server infrastructure for potential deployment or further testing, selected ones of the remaining individuals from the pool of candidate individuals.

### Claim 38 (depends on 37)

The method according to claim 37, wherein the step of storing comprises the step of storing, in association with each of the candidate individuals in the testing subset, the respective first history so as to identify both the data samples on which the respective individual has been tested and the performance of the respective individual on each of such data samples.

### Claim 39 (depends on 38)

The method according to claim 38, wherein the step of the computer system selecting individuals for discarding in dependence upon comparisons among their respective overall fitness estimates, selects the individuals for discarding in further dependence upon a characteristic of the individuals' first histories other than their respective overall fitness estimates.

### Claim 40 (depends on 37)

The method according to claim 37, wherein the step of the computer system testing each individual in a testing subset comprises the steps of:
the computer system identifying, in dependence upon the first history associated with a given individual, a given data sample not already identified in the first history associated with the given individual;
retrieving the given data sample from the data mining training database; and
testing the given individual on the given data sample.

### Claim 41 (depends on 37)

The method according to claim 37, wherein the step of the computer system testing each individual in a testing subset comprises the steps of:
the computer system receiving a block of at least one data sample;
determining in dependence upon the first history associated with a given individual whether each data sample in the block is already identified in the first history associated with the given individual; and
testing the given individual on only those data samples in the block, if any, which are not already identified in the first history associated with the given individual.

### Claim 42 (depends on 27)

The method according to claim 27, wherein the step of the computer system testing each individual in a testing subset comprises the steps of:
the computer system identifying, in dependence upon the first history associated with a given individual, a given data sample not already identified in the first history associated with the given individual;
retrieving the given data sample from the data mining training database; and
testing the given individual on the given data sample.

### Claim 43 (depends on 27)

The method according to claim 27, wherein the step of the computer system testing each individual in a testing subset comprises the steps of:
receiving a block of at least one data sample;
determining in dependence upon the first history associated with a given individual whether each data sample in the block is already identified in the first history associated with the given individual; and
testing the given individual on only those data samples in the block, if any, which are not already identified in the first history associated with the given individual.

### Claim 44 (depends on 29)

The method according to claim 29, wherein the steps of determining a union and forming a combined history collectively comprise the step of, for each particular data sample identified in the further history:
determining whether the particular data sample is already identified in the first history associated with the particular individual; and
adding an identification of the particular data sample to a history associated with the particular individual only if the particular data sample is not already identified in the first history associated with the particular individual.

### Claim 45 (depends on 32)

The method according to claim 32, wherein the step of the computer system selecting individuals for discarding in dependence upon comparisons among their respective overall fitness estimates, selects the individuals for discarding in further dependence upon a characteristic of the individuals' first histories other than their respective overall fitness estimates.

### Claim 46 (depends on 32)

The method according to claim 32, wherein the step of calculating an overall fitness estimate for each of the individuals in the testing subset comprises the step of calculating the overall fitness estimate for each particular one of the individuals in the testing subset in further dependence upon a characteristic of the first history associated with the particular individual other than an un-weighted average performance of the particular individual.

### Claim 47 (depends on 32)

The method according to claim 32, wherein the step of calculating an overall fitness estimate for each of the individuals in the testing subset comprises the steps of:
calculating a preliminary fitness estimate for each particular one of the individuals in the testing subset; and
weighting the preliminary fitness estimate for at least some of the individuals in the testing subset in dependence upon an additional characteristic of the first history associated with the particular individual.

### Claim 48 (depends on 47)

The method according to claim 47, wherein the step of weighting the preliminary fitness estimate for at least some of the individuals in the testing subset comprises the step of weighting the preliminary fitness estimate for only those of the individuals in the testing subset which have been tested on at least a predetermined number N>1 of data samples.

### Claim 49 (depends on 47)

The method according to claim 47, further comprising the step of providing in the memory data identifying layer parameters for a number T>0 of gene pool experience layers L1-LT in an elitist pool, the layer parameters for each i'th one of the layers L1-LT identifying a respective minimum testing experience level ExpMin(Li) for the layer, ExpMin(L1)>1 and, if T>1, ExpMin(Li)>ExpMin(Li−1) for each i>1,
and wherein the step of weighting the preliminary fitness estimate for at least some of the individuals in the testing subset comprises the step of weighting the preliminary fitness estimate for only those of the individuals in the testing subset which have been tested on at least a predetermined number N of data samples, where N is equal to one of the ExpMin(Li).

### Claim 50 (depends on 32)

The method according to claim 32, wherein the step of the computer system selecting individuals for discarding in dependence upon comparisons among their respective overall fitness estimates, selects the individuals for discarding in further dependence upon a comparison of a particular characteristic of one individual's first history other than its overall fitness estimate, with the particular characteristic of a different individual's first history.

### Claim 51 (depends on 32)

The method according to claim 32, wherein the step of providing for deployment of selected ones of the remaining individuals from the pool of candidate individuals, comprises the step of selecting for deployment individuals from the pool of candidate individuals in dependence upon an additional characteristic of the first histories associated with the individuals in the pool of candidate individuals.

### Claim 52 (depends on 27)

The method according to claim 27, further comprising, after the step of selecting and before the step of providing for deployment, a procreation step of forming new individuals in the pool at least in part by copying into each subject new individual at least one member of the group consisting of: a condition in an individual in a set of at least one parent individual corresponding to the subject new individual; and an output in an individual in the set of parent individuals corresponding to the subject new individual,
and further comprising, after the procreation step and before the step of providing for deployment, repeating the steps of testing, storing, calculating and selecting at least once.

### Claim 53 (depends on 27)

The method according to claim 27, wherein each of the data samples in the training database indicates historical market data at a plurality of times in a trading day, including the price of a security at the plurality of times,
wherein each individual indicates a plurality of conditions resolvable in dependence upon the data in a given one of the samples, the outputs identified by the individuals including security trading instructions,
wherein the step of testing comprises the step of, for each trial of a given one of the individuals being tested, hypothetically performing all the trading instructions made by the given individual on a respective given one of the samples throughout the trading day of the given sample,
and wherein the step of calculating an overall fitness estimate comprises the step of updating the fitness estimate associated with the given individual in dependence upon the given individual's profit or loss for the trading day of the given sample.