Great video. I have been experimenting with ML. It's a lot to take in for an individual trader, getting the data cleaned and fixing errors in the model has kept me from making much progress there. This system is very user friendly.
Keep experimenting! That's what the Lab is all about. Also, if you'd like to join us, we're hosting an event tomorrow where we'll be going over concepts for refining models. Here's a link to sign up: us06web.zoom.us/webinar/register/WN__hV-COEKRyOpwlHhFH48yg
Excellent video! Regarding bot entries and exits, will the bot exit via take profit, stop loss or candles passed if the entry criteria is still valid? If not, backtesting for bot trading is invalid, and for bot trading, these results are misleading. Or are AI strategy 'signal emerged' entries treated differently? Also, how did you determine the new adjusted TP, SL, and candles passed parameters? Isn't this overfitting? Thanks!
Bots will execute exits the same way a strategy test does-based on whichever exit condition (take profit, stop loss, or candles passed) is hit first. This ensures consistency between backtesting and live bot trading. Regarding the adjusted TP/SL parameters, we used the example section of the chart to refine these settings. Multiple iterations were tested before settling on the 5%/2%/35-candle window, which we felt struck a good balance. This process involved some discretionary judgment, but the adjustments were data-informed. As for concerns about overfitting, it's a valid consideration. To address this, forward testing would be the logical next step to validate the strategy on future trades in a live environment. Additionally, applying the strategy to an asset it wasn’t trained on (like we did here with the $QQQ) can help assess its robustness and reduce the likelihood of overfitting.
@@TrendSpiderUnless AI strategies are different, or something has changed, bots cannot be backtested, as they will not exit while an entry criteria is valid. Strategy testing can and will exit in this case, bot testing will not. There is therefore no consistency between strategy testing and live bot trading.
@@haverelmink Bots are built from backtests, so by nature they must be backtested in order to function. A bot is really more of a "forward test" where the bot logic is run forward. You cannot backtest into the future, but that's OK because you can run a bot without a broker connection on it, soy ou can see what it would do in real time before connecting it to an account.
@@du1081TrendSpider strategy testing functions differently from bot behavior. Bots cannot be currently backtested unless entry and exit criteria are mutually exclusive. Forward testing is the only way to test bots.
@@du1081TrendSpider bots cannot be backtested, as bots will not exit a position while entry criteria is valid. Strategy testing can and does exit with valid simultaneous entry and exit conditions. This is not the logic that bots follow, so the strategy tester cannot be used for bot backtesting. Bots can only be forward tested.
Thanks this video explained a lot more. Can you explain "Confidence" a little more. Most of the strategies ive tested are a little better with a lower Confidence than high confidence.
Confidence is simply how confident the model is that a signal will produce a move that will result in your TP hitting before your SL. With a lower confidence setting, you're likely to take more trades, which could be the reason why you're seeing better results. It's really all a matter of fine tuning in the strategy tester to find what works best!
Backtest deta result alwas diffrent than live market. Why dont do live trade in live market. Why all stratji run in past market deta, why dont in live market??
Great video. I have been experimenting with ML. It's a lot to take in for an individual trader, getting the data cleaned and fixing errors in the model has kept me from making much progress there. This system is very user friendly.
Keep experimenting! That's what the Lab is all about. Also, if you'd like to join us, we're hosting an event tomorrow where we'll be going over concepts for refining models. Here's a link to sign up: us06web.zoom.us/webinar/register/WN__hV-COEKRyOpwlHhFH48yg
Excellent video! Regarding bot entries and exits, will the bot exit via take profit, stop loss or candles passed if the entry criteria is still valid? If not, backtesting for bot trading is invalid, and for bot trading, these results are misleading. Or are AI strategy 'signal emerged' entries treated differently?
Also, how did you determine the new adjusted TP, SL, and candles passed parameters? Isn't this overfitting?
Thanks!
Bots will execute exits the same way a strategy test does-based on whichever exit condition (take profit, stop loss, or candles passed) is hit first. This ensures consistency between backtesting and live bot trading.
Regarding the adjusted TP/SL parameters, we used the example section of the chart to refine these settings. Multiple iterations were tested before settling on the 5%/2%/35-candle window, which we felt struck a good balance. This process involved some discretionary judgment, but the adjustments were data-informed.
As for concerns about overfitting, it's a valid consideration. To address this, forward testing would be the logical next step to validate the strategy on future trades in a live environment. Additionally, applying the strategy to an asset it wasn’t trained on (like we did here with the $QQQ) can help assess its robustness and reduce the likelihood of overfitting.
@@TrendSpiderUnless AI strategies are different, or something has changed, bots cannot be backtested, as they will not exit while an entry criteria is valid. Strategy testing can and will exit in this case, bot testing will not. There is therefore no consistency between strategy testing and live bot trading.
@@haverelmink Bots are built from backtests, so by nature they must be backtested in order to function. A bot is really more of a "forward test" where the bot logic is run forward. You cannot backtest into the future, but that's OK because you can run a bot without a broker connection on it, soy ou can see what it would do in real time before connecting it to an account.
@@du1081TrendSpider strategy testing functions differently from bot behavior. Bots cannot be currently backtested unless entry and exit criteria are mutually exclusive. Forward testing is the only way to test bots.
@@du1081TrendSpider bots cannot be backtested, as bots will not exit a position while entry criteria is valid. Strategy testing can and does exit with valid simultaneous entry and exit conditions. This is not the logic that bots follow, so the strategy tester cannot be used for bot backtesting. Bots can only be forward tested.
Thanks this video explained a lot more. Can you explain "Confidence" a little more. Most of the strategies ive tested are a little better with a lower Confidence than high confidence.
Confidence is simply how confident the model is that a signal will produce a move that will result in your TP hitting before your SL. With a lower confidence setting, you're likely to take more trades, which could be the reason why you're seeing better results. It's really all a matter of fine tuning in the strategy tester to find what works best!
What do you think of Tickeron.
does this work with NQ, ES and RTY?
Yes! Works with all US equities, ETFs, OTCs, Futures, Crypto and FX
Hey @marketrhythms, to echo what @du1081 said here; Yes! It works w/ all US equities, ETFs, OTCs, Futures, Crypto, and FX.
i dont see rsi model 2 as an available indicator choice
A model is only available in the indicators once it's deployed. Deployed models cannot be shared yet, but that's coming!
Backtest deta result alwas diffrent than live market. Why dont do live trade in live market. Why all stratji run in past market deta, why dont in live market??