ValTicker
A good idea. Did it ever pay?
Buy high dividends, buy low P/E, follow the big investors: there are many ways to pick stocks. The real question is one: done this way, how did it turn out over the past decade and more?
- Try an idea on past data first.
- A good-looking backtest can be luck.
- Retest it; keep only what holds up.
Backtest: try it before you put money in
A backtest runs your idea on real past data, from start to finish: pick, rebalance, pay the trading costs, then see the return, the max drawdown and every year on its own. See the result before any money goes in.
A good number is not always a true one
Trying hundreds of conditions and keeping the best one; using data no one had at the time; leaving out trading costs. Each makes a backtest look better than reality. That is why screening comes next.
Screening: was it luck?
Test the same strategy again with other periods, other stocks and higher costs. If it still holds up, the conditions work; it was not just good timing.
- Does it still hold with higher trading costs?Higher costBacktests again with trading costs several times higher
- Does it still hold with similar stocks swapped in?Peer swapSwaps each stock on the list for another stock that has performed closely with it, and backtests again
- Does it rely on just one period?SegmentsSplits the backtest period and looks at each part on its own
- Does it rely on just one half of the stock pool?Universe halvesSplits the stock pool in two and backtests each half
- Does it still hold on a random half of the stocks?Random half universeReruns on a random half of the stocks, many times
- Do the rules still work on the data that comes next?Walk-forwardSets the rules on one stretch of data, checks them on the next, and moves forward
- Does the strategy beat randomly picked lists?PlaceboRandom lists with the same number of stocks and rebalance timing, side by side with the strategy
- After many strategies were tried, is the result just luck?DSRA Sharpe ratio check that counts how many strategies were searched
ValTicker
Tell Claude your stock-picking idea. ValTicker builds strategies from it, backtests and validates them, and narrows them down to the one that's yours.
- Tell Claude your idea
- builds strategies
- backtests and validates
- narrows them down
For professional users
Look-ahead cannot be written. Stocks are picked after the close and bought at the next open, with costs fixed by the engine. The same idea gives the same answer every time. DSR is a Sharpe ratio check that counts how many strategies were searched.
Ask one question and see how your idea did.
