Lately, my backtest results over various strategies, datasets, and ranges have been literally, unbelievably good. The problem seems to be in both “monthly returns” and "periodic returns”. To illustrate, I ran the simple SMA50/200 crossover strategy on a dataset of leveraged, inverse ETFs, which should be hard or impossible to show profitable results using a simple long only strategy. I am attaching a pdf of the excel spreadsheet generated by the strategy results.
Please advise on how to correct.
7999-SMA50-200-crossover-pdf
Please advise on how to correct.
7999-SMA50-200-crossover-pdf
Rename
Periodic returns measure changes in the equity curve at different intervals.
That's a far different measure than "closed" position returns. Apples and oranges.
That's a far different measure than "closed" position returns. Apples and oranges.
Net total of profits and losses reflected in closed positions over any interval (t_0 to t-1) is the cash portion of the change in the equity curve between t_0 and t_1. When there are no open positions held by a strategy the market portion of the equity curve between t_0 and t_1 does not change, so the cash portion is the total portfolio value at t_1. In the illustration I sent, the strategy had no open positions after 6/17/2024, but the periodic returns report continued to show gains, or changes to the equity curve, after 6/17.
But the daily returns are showing gains every day after what is shown as the last exit. I can only guess that it's cash interest applied to the backtest, but if it were only that I'd expect the daily gain to have less variance. There's something else going on that you're not showing us.
If you can't figure it out, try posting images of the WealthLab's Positions and Metric reports, and also the Backtest Preferences.
If you can't figure it out, try posting images of the WealthLab's Positions and Metric reports, and also the Backtest Preferences.
I think you are right – cash interest received seems to be the difference. I inadvertently set the initial capital at $4,000,000 rather than $400,000. I can’t (or at least didn’t) calculate the interest contribution to portfolio value over the last, most recent period, but given the sizable difference in starting capital, interest probably accounts for the entire difference between net gain shown by “positions” and net gain in portfolio value reflected in “periodic returns”.
You can disable interest (and dividend) payments in the Backtest Preferences (F12).
Thanks. I have done that for my"trading" vs. "investing" strategies. I thought I had already done that, but somehow it must have gotten modified.
Your Response
Post
Edit Post
Login is required