Support Board
Date/Time: Mon, 25 Nov 2024 02:58:14 +0000
Post From: Sierra Charts Backtester (PssSCBT)
| [2019-08-29 16:53:34] |
| User565689 - Posts: 19 |
|
Hello JayShea, I have a bad habit of not updating the instances of Sierra Chart that I use for SCBT that is a carry over from the days of corporate Windows updates always breaking critical functionality (yes, I know that is terrible security practice, but it's not fun to have a critical system break right in the middle of a product launch or the trading day, depending on your field). I personally am still on SC v1875 for my backtests since I have not needed any of the new updates from SC. In the old days (and even with SC/SCBT) it was always risky to roll out a new Windows update without thorough testing, and it hasn't been worth the effort to update and make sure nothing was broken for me at this time. With the exception of the major Trade Activity Log breaking update in early 2019 (which I reference in my review and Burkhard worked with me to make SCBT work with SC again after that major SC change), it was normally a painless process to update SC to each new version, but I have not had the time to try since then. If there is another code-breaking change like that one, then I assume Burkhard would be willing to tweak the code again to make it work, but since I have no affiliation with him I can't say for sure if SCBT is still under active maintenance. As for your question about speed, SCBT uses the native SC backtesting process, so it is almost as fast as SC's backtests. I say almost since SCBT adds some overhead as it changes Inputs, saves the Trade Activity Log, reloads the chart, and the like. Those are all unavoidable actions if you want to iterate parameters, though, so you can't fault SCBT for that. Once the parameters are entered and the chart book is loaded, it runs at 100% of the SC backtest speed, which is limited by your computer. I run it in "Accurate Trading System Back Test Mode", which can take quite a while depending on how involved your custom studies are, but SCBT also supports the Bar Based Back Test Mode (Standard Replay), which runs very quickly. The downside to Standard Replay is that it only uses the OHLC data instead of tick data, so it is not that accurate if you make trades (entry or exit) within a bar instead of just on bar close. I have not seen any reduction in speed just due to SCBT. For example, if a backtest in SC takes 1 hour (after inputs are changed, etc.), then that same backtest using SCBT to control SC will also take 1 hour. You just have to add a few seconds to a minute extra for each backtest iteration to allow SCBT to do all of its supervisory functions. |