backtesting / algotrading / walk forward testing
Backtesting vs Data Analysis
If you want to trade like the people on Wall Street, you need to understand the difference between backtesting and data analysis. A backtest is a historical simulation of how a set of trading rules performed, with results like percent gain, Sharpe ratio and Sortino ratio. Data analysis on raw data can't do that. A query can't rebalance the top 10 stocks by market cap every 14 days with a stop loss, handle mergers, acquisitions and ticker changes, model limit orders, or become a live trading engine. Account for slippage, walk-forward optimization and dividends too.
Transcript
0:00If you're trying to trade stocks like the billionaires that work on
0:03Wall Street then you need to understand the difference between backtesting and data analysis
0:09Here's the key idea When you run a backtest you're performing a historical simulation of how a set of trading rules performed in the past
0:20A backtest might look something like this where you have a strategy a set of trading rules and then you see the historical results of these
0:32rules You can see things like the percent gain the
0:35Sharpe ratio
0:37the Sortino ratio This is how this set of rules performed across this period in the past
0:44This is different than data analysis because you need a simulation engine
0:49You can't just analyze the raw data Unless your strategy is extraordinarily simple like buy and hold or double down every 7 days then you need
0:59to write code to codify your trading rules For example you can't just write a query to create a trading strategy that rebalances the top 10
1:09stocks by market cap every 14 days and include a stop loss for that strategy
1:16You can't make it account for merger changing their names
1:24You can't make it account for limit orders And most importantly you can't use that simulation engine to deploy a live trading engine to execute trades
1:35in real time
1:37They're so distinctly fundamentally different that if I'm wrong I just want you to prove it
1:43Let me see the queries you write to get the same results as some of these strategies
1:49This is an easy one so you should have really no trouble at all
Join the conversation
Loading conversation…