How does wavealgo decide whether a Pine script has alpha?

Every script is executed by the wavealgo JS engine across a fixed grid of 6 markets × 3 timeframes (18 cells), with a 0.05% per-side fee and market orders filling at the next bar open. A script that reads unconfirmed values is rejected before performance is measured at all, because its historical results cannot be reproduced live.

What do the Alpha Desk verdicts mean?

Verified means positive out-of-sample expectancy that survives costs. Conditional means the edge exists only in a specific regime or role. Repaints means the script reads information unavailable at signal time, so its history cannot be trusted. No alpha means it was measurable and did not beat its costs.

How many Pine scripts has wavealgo audited?

20 scripts have completed audit — the v1 reference catalog of classic public techniques. The Alpha Desk shows every one of them, ranked by alpha score from 0 to 100.

Pine Script leaderboard — every audit, ranked

/ to focus

#	Script	Verdict	Alpha	Sharpe	Max DD	PF	Trades
01	Golden Cross (SMA 50/200) @wavealgo	Conditional	49	0.78	-32.5	1.49	765
02	Keltner Breakout (20, 2) @wavealgo	Conditional	44	0.71	-25.8	1.32	2845
03	EMA Cross (9/21) @wavealgo	Conditional	42	0.73	-39.7	1.14	5696
04	ATR Channel Breakout (20, 2×ATR14) @wavealgo	Conditional	42	0.58	-31.3	1.18	3544
05	Donchian Breakout (20/10) @wavealgo	Conditional	40	0.63	-36.5	1.24	5832
06	Chandelier Exit Trend (22, 3) @wavealgo	Conditional	40	0.65	-38.0	1.19	4185
07	Bollinger Breakout (20, 2) @wavealgo	Conditional	39	0.62	-33.3	1.15	4740
08	MACD Cross (12/26/9) @wavealgo	Conditional	37	0.47	-43.0	1.09	9907
09	Ichimoku Conversion/Base Cross @wavealgo	Conditional	37	0.54	-32.1	1.18	3255
10	RSI Mean Reversion (14, 30/70) @wavealgo	Conditional	35	0.39	-41.3	1.30	1023
11	Stochastic Cross (14,3,3) @wavealgo	No alpha	34	0.28	-44.3	1.00	4365
12	RSI 50-Line Momentum (14) @wavealgo	Conditional	34	0.35	-49.2	1.18	14399
13	Supertrend Follow (10, 3) @wavealgo	No alpha	33	0.31	-45.7	1.05	3060
14	Parabolic SAR Flip (0.02, 0.2) @wavealgo	Conditional	33	0.36	-48.2	1.07	10508
15	Williams %R Reversal (14) @wavealgo	Conditional	29	0.23	-43.8	1.01	6204
16	MFI Reversion (14, 20/80) @wavealgo	Conditional	29	0.23	-39.0	1.30	1007
17	ADX-Filtered DI Cross (14, 20) @wavealgo	Conditional	28	0.26	-36.1	1.12	4320
18	Bollinger Reversion (20, 2) @wavealgo	No alpha	26	0.14	-41.7	0.93	4824
19	CCI Reversion (20, ±100) @wavealgo	No alpha	25	0.10	-44.5	0.94	5142
20	OBV MA Cross (20) @wavealgo	Conditional	23	0.05	-41.0	1.11	16265

Audit · #0001

Golden Cross (SMA 50/200)

@wavealgo·Strategy Conditional

49/ 100 alpha

Audit rank
#1 of 20

Full audit page ↗

Verdict rationale

Composite Sharpe 0.78 across 18 cells. The edge concentrates in SPX500 1H (Sharpe 1.82); 1 of 18 cells are negative after the 0.05% per-side fee. Trails buy-and-hold on average.

Alpha score breakdown

Out-of-sample edge

+16

Regime stability

+24

Flaw penalty

Reproduction accuracy

Composite measurements

Sharpe

0.78

Max drawdown

-32.5%

Profit factor

1.49

Win rate

55%765 trades

Audit log

08-08 04:53Transpile OK · pine2js · 20/20 catalog scripts

08-08 04:5318-cell run · 6 markets × 3 timeframes

08-08 04:53Fee overlay · 0.05% per side on every fill

08-08 04:53Verdict filed · Conditional · alpha 49

Measured source · Pine v5

1//@version=5

2strategy("Golden Cross (SMA 50/200)", initial_capital=100000, default_qty_type=strategy.percent_of_equity, default_qty_value=100)

3fast = ta.sma(close, 50)

4slow = ta.sma(close, 200)

5longSig = ta.crossover(fast, slow)

6exitSig = ta.crossunder(fast, slow)

7if longSig

8 strategy.entry("L", strategy.long)

9if exitSig

10 strategy.close("L")

11plot(strategy.equity, "equity")

12plot(strategy.position_size, "pos")

13plot(strategy.closedtrades, "closed")

14plot(strategy.wintrades, "wins")

15plot(strategy.grossprofit, "gp")

16plot(strategy.grossloss, "gl")

Flaw checklist

RepaintingNone · confirmed bars

Look-ahead biasNone · next-bar-open fills

request.securityNot used

Sizing100% equity · long-only

Fee model0.05% per side

SourceOpen · shown left

Reproduction accuracy

This exact source was executed by the wavealgo JS engine — there is no port step to drift. The engine itself is oracle-verified against the Python reference on 251/257 corpus scripts (bar-by-bar, 1e-9 tolerance).

Composite equity · measured window

Mean fee-adjusted equity across the 18 cells (normalized)

By market · timeframe (18 cells)

Symbol · TF	Sharpe	Return	Trades
BTCUSD 1H	0.70	+1155.8%	257
BTCUSD 4H	0.92	+3580.6%	57
BTCUSD 1D	0.60	+629.7%	8
ETHUSD 1H	0.74	+2328.2%	245
ETHUSD 4H	0.99	+12918.5%	49
ETHUSD 1D	0.56	+604.2%	8
SPX500 1H	1.82	+64.2%	10
SPX500 4H	0.85	+19.2%	4
SPX500 1D	0.66	+255.9%	6
QQQ 1H	1.54	+78.8%	10
QQQ 4H	0.81	+36.5%	2
QQQ 1D	0.88	+781.4%	8
GOLD 1H	1.18	+72.0%	14
GOLD 4H	1.47	+109.1%	2
GOLD 1D	0.37	+83.5%	11
EURUSD 1H	0.08	+0.9%	52
EURUSD 4H	0.21	+2.8%	10
EURUSD 1D	-0.11	-9.3%	12

Cost model

0.05% commission per side charged at every fill, market orders fill at the next bar open (TradingView rule), equity is compounded at 1×, slippage is not modeled in v1. Window: whatever history the source provides per cell — daily cells run 3,278–4,000 bars, intraday far shorter for Yahoo-sourced symbols (capped at 730 days by the source), so cells are not equal-length and the composite is a median, not a mean.

Parameter sensitivity

Not yet measured for the v1 reference catalog. Every script here runs the canonical textbook parameters — nothing was tuned, so there is no in-sample optimum to inspect. The sweep (an 11×11 grid per parameter pair, out-of-sample Sharpe per cell) ships with the first third-party submissions, where curve fitting is the thing to catch.

What it will show

Each cell will be the out-of-sample Sharpe of one parameter combination. A sound strategy produces a broad plateau; a curve-fitted one lights up a single point.

↑↓Move /Search 1-4Tabs ⌘KCommand ESCDismiss 20 / 20 shown · v1 reference catalog Sort Alpha ↓ Scored UTC 2026-08-08 04:53