An idea today. The past to test it.
Would your trading idea have held up? Describe the rules in plain language, set the fees and leverage it would really pay, and read the whole result rather than the final number.
- Adjustable fees and leverage
- 20+ metrics
- Trade-by-trade detail
Total return
18.4%
Max. drawdown
-8.2%
Trades
64
Test buying the S&P 500 every month for the last five years.
Understand the journey, not just the return
A backtest makes your assumptions visible: what it cost to trade, how deep the drawdowns went, and how often the edge actually showed up.
Set what trading really costs
Commission as a percentage, a flat amount or per contract; spread; overnight swap fees on long and short; leverage; starting capital and the exact window. Per asset, not one global assumption.
Read 20+ metrics, not one
CAGR and total return, maximum and average drawdown, Sharpe and Sortino, win rate, profit factor, expectancy, SQN, reward-to-risk, time in market, maximum exposure, average position size and the fees paid.
Change one variable, not the code
An agent exposes its thresholds as launch variables: override them per run, or sweep a range and compare the outcomes side by side.
Your hypothesis. A clearer answer.
- 01
Describe the rules
Tell Obside what to buy or sell, under which conditions, and on which assets. It writes the agent; you review the logic.
- 02
Set the window and the costs
Pick the period and starting capital, then the commission, spread, swap fees and leverage each asset should pay.
- 03
Inspect and iterate
Read the equity curve, the drawdowns and every trade, adjust one variable, and run it again.
A few things to know.
Do I need to write code?
No. You type what you want in plain English, and Obside writes the agent for you. You read the rules it produced and press run; the code is there if you ever want to look.
Can I model my broker's fees?
Yes, asset by asset: commission as a percentage, a flat amount or per contract, plus spread, overnight swap fees on long and short, and leverage. Change one, run again, and you see exactly what the costs were worth.
Does a good backtest guarantee future returns?
No, and nobody's does. It is a simulation on past prices: it cannot reproduce every live condition, and the engine ignores dividends. Treat it as evidence about an idea, not a forecast.
Give your next idea a proper test
Set the costs it would really pay, then let the metrics guide your next question.
Test my first ideaKeep exploring Obside.
One workspace. More ways to understand and act on the markets.