ALPHAFINDAlgorithmic Research Platform
Join the waitlist
Early access · In development

From trading idea to
trading live.

Build trading strategies with an AI partner. GPU-accelerated, local, off the cloud.
Define the hypothesis. Set the rules.
Structured research starts with a structured question. Define your hypothesis, set acceptance criteria, and lock in your holdout before testing.
AlphaFind·Science
01Hypothesisname the project, state the claim, declare what would prove it wrong
Tag · family
Inspiration
02Protocollock the experimental design before any trial fires — universe, method, partitiontotal · 5,040 days · 20.0 yrs
2.1Data Universe
Add ticker…AAPL MSFT NVDA AMZN GOOGL META BRK.B LLY TSLA AVGO
Time range
20052010201520202025
2.2Validation Method
Single-split
Sequential walk
Cross-validation
Resampling
In-sampleEmbargoOut-of-sampleSealed holdout
In-Sample 1In-Sample 2In-Sample 3In-Sample 4Holdout20052010201520202025

Diagram shows 4 folds for clarity. After the walk-forward sweep, the strategy is fit on all in-sample data and evaluated a single time on the sealed holdout.

2.3Data Partitioning
In-sample window10y
Out-of-sample test11mo
Step1y
Embargo1mo
Holdout period30%
70%In-sample + Out-of-sample
30%Holdout period
opened once
single-shot OOS evaluation
20052007200920112013201520172019202120232025
In-sample14.0yrs
Out-of-sample—yrs
Holdout6.0yrs
All checks pass — ready to lock.
03Predictionsyour bar for success — pick metrics, write formulas, set tests; whatever you need to convince yourself the hypothesis holds
Start from a template
RecentMean-reversion v3Hard-science (mine)Sharpe-strictPinnedTrend-followingRisk-firstBlank
Acceptance criteria
add another:
Gate scored · Track logged

Pass/fail is set automatically once the experiment runs.

04Experimentthe run, fingerprinted — locked design meets data, sealed under one receiptRUN COMPLETED · 8m 14s

In-Sample fold complete Out-of-Sample fold complete Holdout fold complete · sealed

Fired 2026-04-27 14:32 UTC·Runtime 8m 14s·4,200 trades·924 Out-of-Sample sessions·252 holdout sessions written to vault

Receipt sealed·4 fingerprints

every input fingerprinted — the hashes pin exactly what was sealed and when

Hypothesisa3f24c91e8b27d10
Protocolb914e8d25a3f9c20
Predictions7c8ef015d23a4b80
Data as of2025-04-21
Measured values
GateTrack
1.621.781.851.661.641.291.421.181.59 vaulted
GateTrack
17.8%21.5%20.2%18.4%18.1%10.1%13.6%9.2%12.7% vaulted
GateTrack
9.8%11.0%7.5%8.9%10.4%12.1%9.4%14.2%11.6% vaulted
GateTrack
—————pooled · 0.012—
Predictions vs observed In-Sample / Out-of-Sample judged at runtime — holdout opens at Step 05

6 of 7 In-Sample / Out-of-Sample gates met 1 missed — Compound Annual Growth Rate · Out-of-Sample

RoleCriterionPartitionPredictedObservedMarginStatus
GateSharpe ratioIn-Sample≥ 1.501.71+0.21Met
GateSharpe ratioOut-of-Sample≥ 1.201.37+0.17Met
GateCompound Annual Growth RateIn-Sample≥ 18%19.2%+1.2Met
GateCompound Annual Growth RateOut-of-Sample≥ 12%11.4%−0.6Missed
GateMaximum DrawdownIn-Sample≤ 12%11.0%−1.0Met
GateMaximum DrawdownOut-of-Sample≤ 18%14.2%−3.8Met
Gatep_value(deflated_sharpe)Out-of-Sample< 0.050.0120.038 spareMet
05Resultthe holdout vault opens — predictions sealed at the start, judged nowVAULT OPENED 2026-04-27 14:34 UTC·3 sealed gates revealed

Holdout vault opened · 3 of 3 gates met

sealed at protocol-lock·opened once·untouched until this moment

Holdout predictions vs observed sealed at protocol-lock — Margin is signed against the bar; vs OOS shows holdout drift from the prior partition
RoleCriterionPartitionPredictedOOSObservedMarginvs OOSStatus
GateSharpe ratioHoldout≥ 1.101.371.34+0.24−0.03Met
GateCompound Annual Growth RateHoldout≥ 10%11.4%10.7%+0.7−0.7Met
GateMaximum DrawdownHoldout≤ 20%14.2%13.4%−6.6−0.8Met
Trajectory across partitions each gated metric's three values plotted across In-Sample, Out-of-Sample, and Holdout — the shape of the line is the summary
Sharpe ratio
2.01.00
1.71
1.37
1.34
In-SampleOut-of-SampleHoldout
retention 0.80 · OOS→Holdout stable
Compound Annual Growth Rate
25%12%0
19.2%
11.4%
10.7%
In-SampleOut-of-SampleHoldout
retention 0.59 · thinned in OOS
Maximum Drawdown
20%10%0
11.0%
14.2%
13.4%
In-SampleOut-of-SampleHoldout
stress in OOS · partial recovery
Tracked metrics showing 15 of 47 — per-window values for the metrics that fired alongside the gated set; texture, not verdicts
TrackAnnualized Volatility14.8%15.1%14.2%15.6%14.9%15.4%14.8%16.2%15.1%15.4%
TrackCalmar ratio2.162.423.012.282.041.722.031.312.181.81
TrackWin rate58.2%61.4%59.8%57.1%58.9%55.3%58.7%52.4%61.2%56.9%
TrackAverage Win5.8%6.2%5.9%6.4%5.7%5.2%6.1%4.8%6.5%5.6%
TrackAverage Loss−2.8%−2.6%−2.9%−2.7%−3.1%−3.2%−2.9%−3.6%−2.8%−3.1%
TrackWin / Loss2.072.382.032.371.841.632.101.332.321.81
TrackExpectancy1.21%1.43%1.36%1.18%1.07%0.84%1.31%0.42%1.59%1.04%
TrackAverage Trade Duration18d17d19d18d20d19d18d21d17d19d
TrackExposure76%81%78%74%79%82%77%85%73%78%
TrackTrades per Year312298324307316301289335294305
TrackBest month9.4%11.2%10.7%8.9%9.6%8.2%10.4%7.1%12.6%9.6%
TrackWorst month−5.8%−4.7%−6.2%−5.1%−6.8%−6.4%−5.3%−8.1%−4.9%−6.2%
TrackInformation ratio0.941.081.030.860.910.710.930.581.040.82
TrackBeta vs SPY0.860.820.880.840.870.910.850.940.830.88
TrackAnnualized Alpha vs SPY8.4%10.1%9.6%7.8%8.1%5.2%7.4%3.8%8.6%6.4%
06Next Stepdecide where this run goes, then seed the follow-upDECIDED 2026-04-27 14:36 UTC

9 of 10 gates met · holdout confirmed (3/3)

parent receipt linked into v2·holdout window consumed · v2 will register a fresh holdout from extended history

Trajectory across partitions each gated metric's three values plotted across In-Sample, Out-of-Sample, and Holdout — the shape of the line is the summary
Sharpe ratio higher better
1.711.371.34
ISOOSHOLD
retention 0.80·OOS→Holdout stable
Compound Annual Growth Rate higher better
19.2%11.4%10.7%
ISOOSHOLD
retention 0.59·thinned in OOS
Maximum Drawdown lower better
11.0%14.2%13.4%
ISOOSHOLD
stress in OOS·partial recovery
Disposition how this run leaves the journal — exactly one selection, recorded into the receipt
Seed next experiment pre-fills Step 01 of v2 with the parent-link and the proposed change
v1v2Mega-cap momentum after earnings beatparent · 0x9af4·c218·e2a1
Changewiden post-event capture window from 20 sessions to 25; remaining Protocol parameters held fixed
Every strategy must earn it. No shortcuts.
Curve-fitting is the silent killer of trading strategies. A structured validation pipeline catches it before you do.
01
Idea
A hypothesis, a pattern, or code pasted from an LLM — turned into something testable.
02
In-Sample Optimisation
Every parameter combination searched, ranked, and stress-tested on historical data.
03
Out-of-Sample Testing
If it only works in-sample, it's probably curve-fit. Out-of-sample tests it on data it's never seen.
04
Holdout Validation
A sealed slice of data, reserved from the start. Strong holdout performance strengthens confidence that the edge is real.
IN-SAMPLEOUT-OF-SAMPLEHOLDOUTSHARPE2.14SHARPE1.86SHARPE1.72
Every indicator you need. Every strategy you write.
A library of 150 built-in indicators shipped in. Your own strategies accumulate alongside them — tagged, versioned, dependency-mapped, local.
AlphaFind·Library
Library
▦All assets237
★Favorites12
◷Recent8
Types
◆Strategies27
▲Indicators168
150 built-in · 18 custom
●Primitives42
Your Tags
#trend-following14
#mean-reversion8
#production5
#sandbox11
+New tag…
Provenance
✎Yours87
⚙Built-in150
⌕IN: ALL⌘K
★ Favorites#trend-following+ filtersort: recent ↓
Your work87 ASSETS
NameTypeVerTagsModified
◆MomentumCrossStrategyv1.12trend-fprod2 min★⋯
◆MeanRevSPYStrategyv0.8mean-rev1 h★⋯
▲CustomVWAPIndicatorv1.0volume3 h★⋯
●CrossoverGatePrimitivev2.1logic1 d★⋯
◆TrendBreakoutStrategyv2.4trend-f2 d★⋯
◆PairsArb_ES_NQStrategyv0.3sandbox4 d★⋯
Built-in150 ASSETS · A–Z
▲ADXIndicator—trend—★⋯
▲AroonIndicator—trend—★⋯
▲ATRIndicator—volatility—★⋯
▲Awesome OscillatorIndicator—momentum—★⋯
▲Bollinger BandsIndicator—volatility—★⋯
▲CCIIndicator—momentum—★⋯
▲Chaikin Money FlowIndicator—volume—★⋯
▲Donchian ChannelsIndicator—volatility—★⋯
▲EMAIndicator—trend—★⋯
▲Fisher TransformIndicator—momentum—★⋯
▲Heikin AshiIndicator—price—★⋯
▲Ichimoku CloudIndicator—trend—★⋯
▲Keltner ChannelsIndicator—volatility—★⋯
▲MACDIndicator—momentum—★⋯
▲MomentumIndicator—momentum—★⋯
▲Money Flow IndexIndicator—volume—★⋯
▲OBVIndicator—volume—★⋯
▲Parabolic SARIndicator—trend—★⋯
▲RSIIndicator—momentum—★⋯
▲SMAIndicator—trend—★⋯
▲StochasticIndicator—momentum—★⋯
●StopLossTrailPrimitive—risk—★⋯
▲SuperTrendIndicator—trend—★⋯
▲VWAPIndicator—volume—★⋯
▲Williams %RIndicator—momentum—★⋯
Strategy · Yours · v1.12 ★
MomentumCross
312 lines·12 versions·edited 2 min ago·from Forge
Open in Forge ⌘O
Duplicate · Rename · Export · Relationships
trend-followingproductionequities+ add tag
Buys when short-term momentum confirms a new uptrend and the momentum oscillator agrees. Holds as long as both stay aligned; exits when trend weakens or momentum rolls over.
Rules · in plain English
EntrySMA(25) crosses above SMA(50), and RSI(14) is above 50.ExitSMA(25) crosses below SMA(50), or RSI falls below 40.InstrNVDA · MSFT · AMZN · daily bars
Depends on · 4 pieces
▲
RSIBUILT-IN
Momentum filter
›
▲
SMABUILT-IN
Crossover trigger
›
●
CrossoverGateYOURS
Fires once per cross
›
~/alphafind/library/strategies/MomentumCross.pyindexed in 12ms·offline
Work the way you think.
Forge keeps up.
Code it, build it visually, or have AI write it for you. Same engine either way. Bar-by-bar backtests in seconds.
AlphaFind·Forge
Python 3.12
1@strategy
2class MomentumCross(Strategy):
3  """Dual-SMA crossover with RSI confirmation."""
4  fast    = param(10, 5..30)
5  slow    = param(20, 10..50)
6  rsi_len = param(14, 7..28)
7  rsi_min = param(65, 30..80)
8
9  def on_bar(self, bar):
10    sf, ss, r = SMA(self.fast), SMA(self.slow), RSI(self.rsi_len)
11    if sf > ss and r > self.rsi_min:
12      self.enter_long()
13    elif sf < ss or r < self.rsi_min:
14      self.exit_position()
15
16  def on_fill(self, fill):
17    log(f"{fill.side} @ {fill.price:.2f}")
18
19  def metadata(self):
20    return {"name": "MomentumCross",
21            "version": "1.0.2"}
Visual builder
Source
Indicator
Filter
Gate
Combiner
Action
ENTRYEXITSMA(10)> SMA(20)RSI(14)> 65ANDEnter longSMA(10)< SMA(20)RSI(14)< 65ORExit position
NVDA · 1D
Gate proximity
RSI(14)
Equity
Bar — / —
No lag. No web engine. Your GPU, unleashed.
A library of 150+ indicators and 70 drawing tools, rendered natively on your GPU. Hits 120 fps on capable hardware.
Your metrics, your graphs.
Your report, your way.
Drag in the charts you want. Pin the metrics you actually use. Save the layout once — every run fills it in for you.
AlphaFind·Reports
Hypothesis H-2026-Q1-ER-MOM
Page 1 / 4
Momentum Breakout v7
Earnings-driven momentum · walk-forward validated
Verdict
Promote → Live Paper Trade
Symbol
AAPL
Timeframe
15m
Bars
175,248
Ticks modelled
41.2M
Quality
99.9%
Spread
2 pts
Start balance
$100,000.00
End balance
$167,205.42
Parameters · strategy inputs 18
Entry5 · signal & confirmation
fast_ema_period12bars
slow_ema_period26bars
signal_threshold_sigma0.35σ
confirmation_bars2
volume_filter_mult1.50× avg
Exit4 · targets & stops
take_profit_atr2.40× ATR
stop_loss_atr1.00× ATR
trailing_stop_atr0.80× ATR
max_hold_minutes360min
Risk3 · position sizing
risk_per_trade_pct1.00%
max_concurrent_positions1
max_daily_loss_pct3.00%
Filters3 · session & regime
session_window09:30→16:00ET
atr_floor0.80ATR(14)
spread_cap_pts3pts
Walk-forward3 · from hypothesis journal
is_window_days250d
oos_window_days60d
selection_rulesharpe_oos>1.40
Run provenance
run idA7F3-0412-1630
hypothesisH-2026-Q1-ER-MOM
config hashe7c4a2f9
locked at2026-04-25 13:58 UTC
git ref4a1c9e2 · main
rendered2026-04-25 14:08 NZT
enginealphafind-sim 2.4.1
seed0x7f3a
market data15m · adjusted
In-Sample · 2018-01 → 2021-124y · n=620
Sharpe
1.92
Net profit
+34.5%
Max DD
-6.2%
PF
2.41
Win rate
61%
RF
5.57
Out-of-Sample · 2022-01 → 2023-122y · n=385
Sharpe
1.61
Net profit
+18.4%
Max DD
-8.4%
PF
1.88
Win rate
57%
RF
2.19
Holdout · 2024-01 → 2024-121y · n=235
Sharpe
1.58
Net profit
+12.1%
Max DD
-7.1%
PF
1.81
Win rate
55%
RF
1.70
Balance · Equity · Drawdown
BalanceEquityLinear regressionDrawdown %
$170k$150k$130k$110k$100k
In-Sample · 4y
Out-of-Sample · 2y
Holdout · 1y
0%-5%-10%
2018201920202021202220232024
Performance metrics · full breakdown
36 metrics · partition × all · journal-bound
MetricIn-SampleOut-of-SampleHoldoutOverall
Performance
Start Balance USD100,000.00134,520.18152,932.95100,000.00
End Balance USD134,520.18152,932.95165,025.39167,205.42
Total Net Profit USD+34,520.18+18,412.77+12,092.44+67,205.42
Gross Profit USD58,917.3134,845.6222,784.09116,547.02
Gross Loss USD-24,397.13-16,432.85-10,691.65-49,341.60
Profit Factor2.412.122.132.36
Recovery Factor5.572.191.704.18
Expected Payoff USD55.6847.8351.4654.20
AHPR1.000871.000721.000751.00081
GHPR1.000841.000671.000701.00077
Risk
Sharpe Ratio1.921.611.581.74
Balance DD Absolute USD2,184.104,812.663,921.004,812.66
Balance DD Maximal %-6.20-8.40-7.10-8.40
Balance DD Relative %-5.85-7.92-6.64-7.92
Equity DD Maximal %-6.48-9.12-7.55-9.12
LR Correlation0.9760.9480.9610.968
LR Standard Error USD1,2411,8121,0451,512
Z-Score-1.82 (trend)-0.94-1.21-1.52
Trades
Total Trades6203852351,240
Short · won %312 · 58%184 · 54%112 · 52%608 · 56%
Long · won %308 · 64%201 · 60%123 · 58%632 · 62%
Profit Trades % total378 · 60.9%218 · 56.6%129 · 54.9%725 · 58.5%
Loss Trades % total242 · 39.1%167 · 43.4%106 · 45.1%515 · 41.5%
Largest profit trade+812.40+624.77+498.12+812.40
Largest loss trade-384.10-512.66-421.88-512.66
Avg profit trade155.86159.84176.63160.75
Avg loss trade-100.82-98.40-100.87-99.81
Max consecutive wins count · $11 · 2,1048 · 1,4787 · 1,31211 · 2,104
Max consecutive losses count · $4 · -8125 · -1,0244 · -7445 · -1,024
Avg consecutive wins / losses2.8 / 1.72.4 / 1.92.3 / 1.92.6 / 1.8
Position · correlations & holding
Correlation (Profits, MFE)0.720.680.650.70
Correlation (Profits, MAE)-0.54-0.51-0.48-0.52
Correlation (MFE, MAE)-0.31-0.29-0.27-0.30
Min holding time00:1500:1500:1500:15
Max holding time04:3005:1504:4505:15
Avg holding time01:4201:3801:5101:44
Behaviour · distributions & per-trade shape
Where the strategy earns, how risk is priced per trade, and how streaks cluster. Axes in $USD. Partition colours match the summary page.
Net profit by hour of day
Eastern · 24 one-hour bins
+$5k+$2.5k$0-$2.5k-$5k
0006121824
Hour · ET
Net profit by weekday
Monday through Friday · session profit
+$5k+$2.5k$0-$2.5k-$5k
MONTUEWEDTHUFRI
Weekday
Net profit by month
Jan → Dec · calendar seasonality
+$5k+$2.5k$0-$2.5k-$5k
JFMAMJJASOND
Month
MFE vs MAE · per-trade excursion
Max favourable × max adverse move · point size ∝ |P/L| · n=1,240
ISOOSHoldoutLossMFE = MAE
$0$200$400$600$800
MFE = MAE · break-even frontier
$0$100$200$300$400$500
MAE · $USD (max adverse excursion)
Consecutive streak distribution
How often wins and losses cluster · 1,266 streaks identified across n=1,240 trades
Winning streaksLosing streaks
Winning streaks751 total · longest 11 wins · $2,104 max run P/L
1 win
34245.5%
2 wins
21528.6%
3 wins
11815.7%
4+ wins
7610.1%
Losing streaks515 total · longest 5 losses · −$1,024 max run P/L
1 loss
29857.9%
2 losses
13225.6%
3 losses
5811.3%
4+ losses
275.2%
0100200300400
Count of streaks
Trade Journal · chronological log
Every trade this run produced. Exportable to CSV or the native .afr format for downstream journal analysis. Partition tag reflects the window each trade fell into.
Trade log · last 35 of 1,240
export · CSV · .afr
#OpenCloseDirSizeEntryExitHoldMFEMAEP/L $P/L %Part
12062024-11-11 10:302024-11-11 12:00LONG500229.62231.8401:30+1,280-94+1,110+0.87H
12072024-11-12 14:002024-11-12 15:15SHORT500232.88232.9201:15+340-260-20-0.02H
12082024-11-13 09:302024-11-13 11:00LONG500231.06233.2201:30+1,280-68+1,080+0.85H
12092024-11-14 10:452024-11-14 12:00LONG500233.84234.1801:15+480-220+170+0.13H
12102024-11-15 13:302024-11-15 14:45SHORT500234.92234.0401:15+620-118+440+0.35H
12112024-11-18 10:002024-11-18 12:00LONG500233.42235.8802:00+1,420-46+1,230+0.98H
12122024-11-19 09:302024-11-19 10:45SHORT500236.22236.4201:15+220-320-100-0.08H
12132024-11-20 10:152024-11-20 11:45LONG500235.68237.4001:30+1,020-88+860+0.68H
12142024-11-21 13:002024-11-21 14:30LONG500237.84239.2201:30+860-72+690+0.55H
12152024-11-22 09:452024-11-22 11:00SHORT500240.18239.3201:15+580-132+430+0.34H
12162024-11-25 10:302024-11-25 12:00LONG500238.92238.1401:30+320-640-390-0.31H
12172024-11-26 10:002024-11-26 11:45LONG500240.34242.1801:45+1,180-62+920+0.73H
12182024-11-27 13:302024-11-27 15:00LONG500242.62243.8801:30+780-80+630+0.50H
12192024-11-29 10:152024-11-29 11:30SHORT500244.44243.6201:15+520-98+410+0.33H
12202024-12-02 10:002024-12-02 11:45LONG500243.18245.4401:45+1,340-50+1,130+0.89H
12212024-12-03 09:302024-12-03 11:00SHORT500245.82245.9401:30+260-280-60-0.05H
12222024-12-04 10:452024-12-04 12:15LONG500244.72246.8801:30+1,260-66+1,080+0.85H
12232024-12-05 13:152024-12-05 14:30LONG500247.18247.5201:15+520-190+170+0.14H
12242024-12-06 10:002024-12-06 11:30SHORT500248.40247.5601:30+640-110+420+0.33H
12252024-12-09 10:302024-12-09 12:00LONG500246.84249.1201:30+1,380-54+1,140+0.92H
12262024-12-10 09:302024-12-10 11:00SHORT500249.88248.9401:30+580-128+470+0.38H
12272024-12-11 14:002024-12-11 15:15LONG500248.22247.8801:15+310-440-170-0.14H
12282024-12-17 10:152024-12-17 12:00LONG500249.14250.8201:45+1,040-72+840+0.67H
12292024-12-18 10:152024-12-18 12:00LONG500249.82251.9801:45+1,280-124+1,080+0.87H
12302024-12-19 09:302024-12-19 10:15SHORT500252.10251.2400:45+520-180+430+0.34H
12312024-12-19 14:002024-12-19 15:30LONG500250.88249.4201:30+210-840-730-0.58H
12322024-12-20 10:452024-12-20 12:30LONG500248.92251.1801:45+1,410-62+1,130+0.91H
12332024-12-23 09:452024-12-23 11:00SHORT500253.22252.1401:15+640-110+540+0.43H
12342024-12-24 10:002024-12-24 11:15LONG500254.18255.6201:15+860-84+720+0.57H
12352024-12-26 14:302024-12-26 16:00LONG500256.40258.0401:30+950-120+820+0.64H
12362024-12-27 09:302024-12-27 10:45SHORT500258.88258.9201:15+320-240-20-0.02H
12372024-12-27 13:152024-12-27 15:00LONG500257.92259.8801:45+1,180-96+980+0.76H
12382024-12-30 10:002024-12-30 11:30LONG500260.14260.4401:30+460-280+150+0.12H
12392024-12-30 14:452024-12-30 16:00SHORT500261.22260.1801:15+720-85+520+0.41H
12402024-12-31 10:302024-12-31 12:15LONG500259.44261.6201:45+1,320-48+1,090+0.85H
Every method you need.
Already wired in.
From grid search to walk-forward validation, AlphaFind gives you the full research stack in one engine.
Search
GRID · 11 × 11S* = 1.845035205lookback0.51.01.52.02.5threshold
Grid Search
Exhaustive parameter sweep
SOBOL (2,3) · n = 187S* = 1.711.00.50.0vol target00.51.0stop-loss
Quasi-Random Sampling
Sobol · LHS · space-filling
GP · MATÉRN-5/2 · iter 18 / 50S* = 1.78μf(θ)EIθ → next sample
Bayesian
Gaussian process surrogate
METROPOLIS · step 4128T = 0.18energyθT(t)
Simulated Annealing
Escapes local optima
Evolutionary
NSGA-II · GEN 24/50PARETO 1030%15%5%drawdown01.02.0sharpe →
NSGA-II
Multi-objective Pareto search
DE/RAND/1/BIN · F=0.5 CR=0.9GEN 181.00.50.0param B00.51.0param Aabcv = a + F(b−c)
Differential Evolution
Mutation-based optimization
PSO · ω=0.7 c1=c2=2.0ITER 32+10-1x₂-10+1x₁g_best
Particle Swarm
Swarm convergence search
CMA-ES · λ=20σ = 0.42+20-2x₂-20+2x₁gen 1gen 8m
CMA-ES
Adaptive covariance strategy
Validation & Stress Testing
WALK-FORWARD · ANCHOREDOOS = 1.32201820212024ISOOSISOOSISOOSISOOSISOOS5 folds · step 6 mo · test 6 mo
Walk-Forward
Anchored IS · rolling OOS
K-FOLD CV · K=5μ OOS = 1.35testfold 1testfold 2testfold 3testfold 4testfold 5S1.321.281.451.181.50
K-Fold CV
Stratified · repeated · grouped
CPCV · N=10 k=2120 pathsπ1π2π3π4π5π6test · purged · train
CPCV
Combinatorial purged CV
BOOTSTRAP · n = 10,000CI 95%1.6×1.0×0.7×equityD0D ½DTtime
Monte Carlo
10,000 bootstrap paths · 95% CI

Be there the moment we open the doors.

Join the waitlist. We’ll email you the day AlphaFind goes live.