Parcourir les catégories
Explorer
Fiverr Pro
Français
$
USD
I will implement and backtest trading strategy in python
Jianshou
À propos de ce service
You describe the rules. I turn them into working Python code and a backtest report you can reproduce on your own machine.
I have spent years building strategies on a live brokerage platform - real capital, real fills, not just notebooks. That is why my reports show the fill assumptions and adjustment methods most freelancers skip.
I work across markets. The same code runs on US equities, ETFs, crypto spot, or any OHLCV file you send - only the cost and settlement settings change between markets, not the strategy.
What you get:
- Commented Python source code
- Full backtest report: parameters, fees, slippage, equity curve, drawdown, win rate
- Raw trade log exported by the framework, not typed by hand
- Parameter table: what each one does, default, useful range
- A note on how to re-run it and which fields to compare
What I do not do:
- No investment advice
- No profit promises. A strategy can be built perfectly and still lose money
- No signals and no ready-made systems
Before you order, send me your entry, exit, sizing and stop rules. If something is ambiguous I will ask. Fixed price, fixed date, no hourly surprises.
Découvrez Jianshou
Jianshou
Quant Backtesting and Strategy Validation in Python
- DeChine
- Membre depuisoct. 2026
Langues
Chinois, Anglais
I am a quantitative developer building trading strategies in Python.
I have run strategies live in China A-shares: real capital, real fills, T+1 settlement, stamp duty on every sell - a stricter rulebook than US equities. That is why my reports spell out fill assumptions and adjustment methods.
Same code runs on US equities, crypto spot, or any OHLCV file. Only cost and settlement settings change.
Not for you if you want investment advice, a promise about returns, or someone who says yes to everything. I will tell you if your idea has a flaw before you pay me.
Mon portfolio
FAQ
Do you guarantee profits?
No. I implement and test what you specify; I do not predict market outcomes. A strategy can be coded perfectly and still lose money - that is a market result, not a coding error. What I do stand behind: the logic matches your specification, and the numbers in my report reproduce on your machine.
What do you need from me to start?
Four things: entry and exit rules, position sizing and stop rules, the market and date range to test, and where your data comes from. A written note is enough - it does not need to be formal.
Which platforms and libraries do you use?
Python: pandas, numpy, matplotlib, backtrader, vectorbt. I also work inside brokerage platforms that run Python - I run my own strategies on one. I do not work with Pine Script (TradingView) or MQL4/MQL5 (MetaTrader).
What exactly gets delivered?
Commented source code, a full backtest report (parameters, fees, slippage, equity curve, drawdown, win rate), the raw trade log exported by the framework, a parameter table, and a short note on how to re-run it.
Can you connect it to my broker for live trading?
That is a larger and separate job: execution, error handling, position tracking, and the failure modes that only show up with real money. Message me and I will quote it separately. I do not attach live trading to a backtest order.
Will you keep my strategy private?
Yes. I do not reuse client logic, do not publish it, and do not discuss projects. If you need a signed NDA before sharing details, send it over and I will sign it before we talk specifics.
Do you only work with Chinese markets?
No. My live trading experience is in China A-shares - T+1 settlement, 100-share lots, daily price limits, stamp duty on every sell - but the engine I deliver is market-agnostic. The same strategy code runs on US equities and ETFs, crypto spot, or any OHLCV file you send. Markets differ in six settin
How do I know the backtest is not look-ahead or overfitted?
Three checks ride along with every delivery. A truncation test re-runs the strategy on the first N bars only - signals must match the full run exactly. A buy-and-hold run must reconcile to shares x last close plus cash within 1e-6. The trade log carries both signal date and fill date, so you can con
Why does the top package cost five times the bottom one?
Different questions, not different amounts of the same thing. Backtest Only: does it run. Stress Audit: is it luck - 20+ parameter runs, a heatmap of where it breaks, costs doubled. Validation: will it work on unseen data - out-of-sample equity, three markets. Fewer files, not less value.
My rules are only half-formed. Can you still help?
Often yes. Tell me what you have and I will point out the gaps. If the logic needs real design work rather than translation, I will say so before you order rather than bill you for guesswork.
