Compatibility 3.3.2

Examples compatibility with the shared EOD downloaders

AlgoQuant 3.3.2 is released. The survey used a pre-release filename; the final release JAR is byte-for-byte identical. Release · Publication and source/checksum verification. No test suites were repeated for promotion.

The survey compiles the entire examples project and exercises each public main entry point against the exact obfuscated customer JAR produced from library commit 1646f3420bff99206f49201c2c8482309f52924f. This checks the artifact a customer uses, including its runtime linkage, rather than substituting library source classes.

• Library development version: 3.3.2-SNAPSHOT (unpublished).
• Customer JAR SHA-256: c120a7c5cdb0180a67319a210381a2eb5827d22794897d32d53d8b779d884f25.
• Examples commit: cbda3909407c549744151359aa05767f12ca8be4.
• Full survey: https://github.com/nmltd/algoquant/actions/runs/39198266116
• Successful app checks: https://github.com/nmltd/algoquant/actions/runs/39198552618

Compile and unit checks

All 84 main Java sources and eight test sources compile with Java 8 against the customer JAR. All 31 JUnit tests pass. These are direct javac/JUnit checks; they are not a claim that the examples Maven lifecycle resolved the unpublished snapshot from a public repository.

The survey exposed a test-order problem: LocalCSI300DataProviderTest could initialize library classes before the shared license was loaded. The examples PR now invokes its existing TestLicense.load() helper in @BeforeClass. The full suite passes with the CSI test first.

Application checks

The synthetic HTTP test passes with 262 NAV observations and 14 executions; the live Yahoo-backed test passes with 253 observations and 11 executions. Both verify report persistence across server restart and exports. The SMA2 and Performance Analyzer Chrome suites pass, covering chart interaction, report navigation/refresh and downloads. The locally built container passes its health and synthetic backtest checks; no container was deployed or published.

The application runner must execute the example unit tests before the browser checks: PerformanceAnalyzerTest generates the JSON/XLSX fixture consumed by performance-browser.cjs.

How to interpret the survey

Each demo runs in an isolated working directory with the repository datasets and configuration linked into it. Swing charts run under Xvfb and are captured as PNGs. Technical demos use their existing synthetic --quick option; other demos use their original main defaults. The CSI300 default runs its two baselines; its opt-in 990-optimizer grid is not part of this survey. Broker/database and other service ports are blocked before authentication or orders. The persistent web server is covered by the dedicated app checks.

Ordinary demos have a three-minute limit. The six explicitly expensive simulations have a 30-minute limit and larger heap; the aggregate RunAllDemo has a 90-second limit because its children are also exercised individually. A timeout is not a completed simulation. A successful main return checks execution and exports, not financial validity or the correctness of every numerical result.

The raw survey initially counted expected Yahoo disk-cache misses as missing required datasets. The corrected classifier removes only the specific INFO-level optional-cache-miss diagnostic before inspecting failures. Logs and successful web fallbacks remain available in the run artifacts. Runtime linkage errors fail the compatibility gate; other failures remain explicitly recorded for review.

Observed issues

• IBStrategyServerDemo needs service credentials and is not an operational broker test.
• MCSimulationDemo hit ArrayIndexOutOfBoundsException: 624 in the numerical Mersenne Twister RNG during the hosted parallel simulation. A local retry against the same refactored JAR completed the calculation, as did ten runs against 3.3.1, before reaching the unavailable local display. This is an unresolved intermittent failure, not a verified passing hosted run. The KST model, HMM, Markov-chain, multinomial and uniform RNG classes in the failure stack are byte-for-byte identical in the two JARs.
• KagiMRBasketSimulationDemo and MultiPeriodKagiMRStrategyDemo returned but logged Yahoo HTTP 400 responses for some stocks in its historical window, including ABNB, APP and COIN for 2003–2015. Its result must not be counted as a clean run.

Reports and per-demo stdout/stderr are attached to the survey run. Customer JARs, compiled classes, credentials and downloaded input caches are not uploaded as new survey artifacts.

• Infantino2010ParamAnalysisDemo returned after Yahoo rejected some requested 2007–2014 stock histories, including CRWD and DDOG.
• CSI300SimulationDemo completed both default baselines in approximately 17 minutes and exported its reports/charts. It also logged missing constituent histories that were skipped and failed CSI300 correlation calculations assigned NaN. This is a completed run with warnings, not a clean numerical result.
• RunAllDemo reached its 90-second aggregate budget. Its individual children were still exercised separately by the survey.

Final outcomes

All 57 entry points are accounted for: 49 clean demo completions, four completions with logged errors/warnings, one failed demo, one credential-blocked demo, one aggregate timeout and one web app covered by successful dedicated checks. No runtime API linkage errors were detected.

The table below uses the corrected classifier. The original artifact JSON retains the earlier false missing-data classifications for optional Yahoo cache misses; per-demo logs are unchanged. A main returning successfully is not a financial/numerical accuracy certification.

Entry point	Outcome	Seconds
RunAllDemo	Timed out	90.15
CandleStickChartDemo	Completed	3.79
MultiChartDemo	Completed	7.63
OHLCChartDemo	Completed	5.41
RangeBarsChartDemo	Completed	2.83
TimeSeriesChartDemo	Completed	3.7
XYSeriesChartDemo	Completed	0.92
SGXTickDataCacheFactoryDemo	Completed	8.57
IBStrategyServerDemo	Credentials required	4.14
BatchParameterizedSimulationDemo	Completed	4.56
BootstrapSimulationSMA2Demo	Completed	18.21
MCSimulationDemo	Failed	6.36
MCSimulationTaskRunnerDemo	Completed	31.12
OtcBrokerSimDemo	Completed	2.23
SensitivityAnalyzerDemo	Completed	3.84
BootstrapSimulationDemo	Completed	67.43
Dai2011LongOnlyDemo	Completed	10.22
KagiMRBasketSimulationDemo	Completed with errors/warnings	150.76
MultiPeriodKagiMRStrategyDemo	Completed with errors/warnings	213.35
Elliott2005StrategyDemo1	Completed	4.78
Elliott2005StrategyDemo2	Completed	3.85
MeanReversionBasedOnSpreadQuantileDemo	Completed	7.32
EconomicModelStrategyDemo	Completed	36.33
FundamentalModelStrategyDemo	Completed	155.84
Infantino2010ParamAnalysisDemo	Completed with errors/warnings	92.93
Infantino2010PredictionAnalysisDemo	Completed	4.42
JanuaryEffectStrategyDemo	Completed	1.98
OULogSpreadCRRASignalDemo	Completed	4.35
GMA12Demo	Completed	1.63
KnightSatchellTran1995MLEDemo	Completed	0.82
Corvalan2005Demo	Completed	1.93
CovarianceComparisonDemo	Completed	77.21
Lai2010NPEBModelDemo	Completed	7.41
MarkowitzBigDemo	Completed	43.54
MarkowitzDemo	Completed	8.16
PerformanceOptimizedYahooEODPortfolioDemo	Completed	10.29
PortfolioOptimComparisons1Demo	Completed	1.63
PortfolioOptimComparisons2Demo	Completed	23.62
PortfolioOptimizationAlgorithmDemo	Completed	0.92
PortfolioOptimizationBacktestDemo	Completed	4.07
SOCPMarkowitzComparisonDemo	Completed	0.62
CoinBacktestDemo	Completed	49.28
CSI300SimulationDemo	Completed with errors/warnings	1039.58
AccumulationDistributionLineDemo	Completed	0.97
ChaikinMoneyFlowDemo	Completed	0.62
SMA2Demo010	Completed	1.03
SMA2Demo011	Completed	0.67
SMA2Demo020	Completed	0.93
SMA2DynamicDemo	Completed	2.03
TutorialStrategyDemo	Completed	2.64
MRModelFixedGridDemo	Completed	1.48
MRModelFixedGridOnSGXDataDemo	Completed	51.32
MRModelPeriodicDemo	Completed	1.48
MRModelRanged1Demo	Completed	2.59
MRModelRanged2Demo	Completed	1.68
MeanReversionPortfolioDemo	Completed	3.09
ToolsWeb	App checks passed	Dedicated job