Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Python backtesting framework for trading strategies. Data feeds, brokers, analyzers, and live trading support. Strategy development with commission models, slippage, and signal-based execution.
.claude/skills/mkurman-backtrader/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-15 | ✗→✓ | ▲ Improved | -24% | 0% |
| case-02 | ✓→✓ | ＝ Same ✓ | -13% | 0% |
| case-03 | ✓→✓ | ＝ Same ✓ | 1% | 0% |
| case-04 | ✓→✓ | ＝ Same ✓ | 10% | 0% |
| case-05 | ✓→✓ | ＝ Same ✓ | -32% | 0% |
Backtrader is a Python backtesting framework for trading strategies. Supports multiple data feeds, live trading, commission/slippage models, custom analyzers, and visualization. Well-suited for equity, futures, and crypto strategy development.
bashuv pip install backtrader
pythonimport backtrader as bt class SmaCross(bt.Strategy): params = dict(short=10, long=30) def __init__(self): sma_short = bt.ind.SMA(self.data.close, period=self.params.short) sma_long = bt.ind.SMA(self.data.close, period=self.params.long) self.crossover = bt.ind.CrossOver(sma_short, sma_long) def next(self): if self.crossover > 0: self.buy() elif self.crossover < 0: self.sell() cerebro = bt.Cerebro() data = bt.feeds.YahooFinanceData(dataname="AAPL", fromdate="2022-01-01", todate="2023-01-01") cerebro.adddata(data) cerebro.addstrategy(SmaCross) cerebro.broker.setcash(10000.0) print(f"Final value: ${cerebro.run()[0]:.2f}") cerebro.plot()
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 17,011 | 12,810 | -25% | 1 | 1 | 0% | 3,167 | 2,903 | -8% | 0 | 0 | — |
case-02 | pass→pass | 7,614 | 4,778 | -37% | 1 | 1 | 0% | 1,369 | 1,190 | -13% | 0 | 0 | — |
case-03 | pass→pass | 13,063 | 10,819 | -17% | 1 | 1 | 0% | 2,461 | 2,482 | +1% | 0 | 0 | — |
case-04 | pass→pass | 13,233 | 11,968 | -10% | 1 | 1 | 0% | 2,458 | 2,695 | +10% | 0 | 0 | — |
case-05 | pass→pass | 4,982 | 2,070 | -58% | 1 | 1 | 0% | 939 | 643 | -32% | 0 | 0 | — |
case-06 | pass→pass | 8,356 | 2,520 | -70% | 1 | 1 | 0% | 1,349 | 754 | -44% | 0 | 0 | — |
case-07 | pass→pass | 10,124 | 4,343 | -57% | 1 | 1 | 0% | 1,847 | 1,190 | -36% | 0 | 0 | — |
case-08 | pass→pass | 7,107 | 2,585 | -64% | 1 | 1 | 0% | 1,194 | 815 | -32% | 0 | 0 | — |
case-09 | pass→pass | 8,463 | 3,741 | -56% | 1 | 1 | 0% | 1,519 | 903 | -41% | 0 | 0 | — |
case-10 | fail→fail | 14,810 | 5,460 | -63% | 1 | 1 | 0% | 2,339 | 1,336 | -43% | 0 | 0 | — |
case-11 | pass→pass | 8,583 | 5,572 | -35% | 1 | 1 | 0% | 1,331 | 1,297 | -3% | 0 | 0 | — |
case-12 | pass→pass | 12,025 | 5,187 | -57% | 1 | 1 | 0% | 1,900 | 863 | -55% | 0 | 0 | — |
case-13 | pass→pass | 10,447 | 3,086 | -70% | 1 | 1 | 0% | 1,770 | 854 | -52% | 0 | 0 | — |
case-14 | pass→pass | 5,344 | 2,869 | -46% | 1 | 1 | 0% | 759 | 812 | +7% | 0 | 0 | — |
case-15 | fail→pass | 15,961 | 8,420 | -47% | 1 | 1 | 0% | 2,556 | 1,954 | -24% | 0 | 0 | — |
case-16 | pass→pass | 9,816 | 4,095 | -58% | 1 | 1 | 0% | 1,707 | 1,060 | -38% | 0 | 0 | — |
case-17 | pass→pass | 4,387 | 2,308 | -47% | 1 | 1 | 0% | 682 | 716 | +5% | 0 | 0 | — |
case-18 | pass→pass | 9,984 | 3,130 | -69% | 1 | 1 | 0% | 1,789 | 949 | -47% | 0 | 0 | — |
case-19 | pass→pass | 8,836 | 5,488 | -38% | 1 | 1 | 0% | 1,623 | 1,357 | -16% | 0 | 0 | — |
case-20 | pass→pass | 13,260 | 4,961 | -63% | 1 | 1 | 0% | 2,367 | 1,302 | -45% | 0 | 0 | — |
case-21 | pass→pass | 8,475 | 3,295 | -61% | 1 | 1 | 0% | 1,541 | 943 | -39% | 0 | 0 | — |
case-22 | pass→pass | 5,959 | 2,174 | -64% | 1 | 1 | 0% | 907 | 678 | -25% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +5 percentage points is the difference between those two pass rates over the 22 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.