Yes. That is exactly it! If one plans on re-calibrating models going forward, the re-calibration needs to be included in the backtest.
Enjoy your vacation!
The missing feature from every back tesing platform
- Thread starter fan27
- Start date
Enjoy your vacation!
Forgive my ignorance, but if the re-calibration of model parameters is an outcome of the backtest then how long of a history should you take into account? Wouldn't this have a possibility of being an instable system?
The case I am arguing for is that new model parameters are really entirely new strategies. My intention is to periodically re-calibrate and select the "best of breed" strategies to trade for the next given period of time. This process would need to be included in the back test. I have not worked out all the details yet but this is my thinking.
fan27,
Will you be using any traditional fixed period/parameter technical indicators, i.e., RSI, MACD, MAs etc, to build strategies? If yes, wont you just be optimizing periods/parameters of those indicators creating new strategies by fitting/applying extra weighting to more recent data?
Am I missing something?
Will you be using any traditional fixed period/parameter technical indicators, i.e., RSI, MACD, MAs etc, to build strategies? If yes, wont you just be optimizing periods/parameters of those indicators creating new strategies by fitting/applying extra weighting to more recent data?
Am I missing something?
Yes. I am using typical indicators among other things. What I do not currently support with my platform is optimizing parameters of a given strategy once it is created (will add that soon). So re-calibrating strategies will involve coming up with a potential brand new group of strategies for each interval, not the same strategies with different parameters. Each new interval will include the most recent data. As for weighting more recent data more, that is not something I have not considered but sounds worthy of research.
"... And I still manually trade new ideas to see if worthy to get programmed. But generally always working on ways to lower drawdowns of intraday and longer term, if you not losing much, profits take care of themselves, all we can really do in trading is control losses and not profits..."
This could be a bumper sticker for traders...
IOW... better entries... Price TA, don't you know... with stop discipline
Go for the gains. Traders should let stops control the risk rather than trying to "filter out the ones that won't work". It makes sense that a trader would want to do this, but the reality is that in filtering out "potentially bad" trades, the trader ends up filtering out waaayyyy more profit than he saves on getting the "filter out" correct.
There are 4 possible outcomes for a trade...
Small gain
Small loss
Big gain
Big loss
The only one the trader can control is "Big loss"... don't allow it to happen! (Optionsellers.com ring a bell with anyone?)
This could be a bumper sticker for traders...
IOW... better entries... Price TA, don't you know... with stop discipline
Go for the gains. Traders should let stops control the risk rather than trying to "filter out the ones that won't work". It makes sense that a trader would want to do this, but the reality is that in filtering out "potentially bad" trades, the trader ends up filtering out waaayyyy more profit than he saves on getting the "filter out" correct.
There are 4 possible outcomes for a trade...
Small gain
Small loss
Big gain
Big loss
The only one the trader can control is "Big loss"... don't allow it to happen! (Optionsellers.com ring a bell with anyone?)
"better entries" as much as I once thought as you do, better entries is a crap shoot of luck whether trading off monthlies or 2 tick charts, we have zero idea if where we get in will be best spot or market makes a small move to go lower and why I average down. Same with gains, we have no idea whether gain will be small or large.
Recent backtesting of new concept I tested has shown for scalping to avoid taking trades with 65% of the time were last trades of normal trading in my past, these are signals after the highs are in uptrend and since my risk is normally greater than rewards, it stands to reason to disallow those signals, and real time shows same results within few percents.
You are more a swing trader? Did you take any buys right after highs were made in February? If you did, they violated "better entries" you referred?
And I am working toward different goals in my life, choosing a different path than just trading my funds, where most OPM seeks lower drawdowns and better returns than Indexes cause much less to make back of drawdowns. Automation often puts out way too many signals, way beyond funds in account, so makes sense to me to filter.
But I do agree, depending on where a trader is in age, younger go for larger profits, but when you get into your sixties, perhaps your mindset may be different.
As always @Scataphagos, like and respect your posts.
"Better entries" = taking a position closer to support/resistance... so that you have both a "low amount to risk" and a "resonable place to stop out as per the charts". (Contrast that notion with "waiting for the MACD cross to confirm the trade".)
I had several buys and sells all through the decline.... all within the concept of "day-swing". The notion of "if you had buys after the high, then you violated the concept of better entries"... is not applicable (unless I screw up, of course.
)I NEVER hold through too big of a decline* (unless one is forced on me), and I NEVER average into a losing position*... which does not preclude "building a position" into what I believe/hope to be a complex top/bottom at significant support/resistance... relatively rare... with stop nearby, of course.
I'm a HUGE advocate of "stop discipline", in case you never noticed.
And hey... if you get sucked into stopping what turned out to be a winning play, you can always reenter.
* Years ago I did. Didn't kill me as the money was smaller then.
Last edited:
As a trader gets older, he should focus even more on not taking any large losses... capital preservation, #1. If he misses a profit opportunity, so be it. He still has his capital intact to play for the next one.
I agree, and hedge every entry so you get even less losses.
"Better entries" = taking a position closer to support/resistance... so that you have both a "low amount to risk" and a "resonable place to stop out as per the charts". (Contrast that notion with "waiting for the MACD cross to confirm the trade".)
I had several buys and sells all through the decline.... all within the concept of "day-swing". The notion of "if you had buys after the high, then you violated the concept of better entries"... is not applicable (unless I screw up, of course.
I NEVER hold through too big of a decline* (unless one is forced on me), and I NEVER average into a losing position*... which does not preclude "building a position" into what I believe/hope to be a complex top/bottom at significant support/resistance... relatively rare... with stop nearby, of course.
I'm a HUGE advocate of "stop discipline", in case you never noticed.
And hey... if you get sucked into stopping what turned out to be a winning play, you can always reenter.
* Years ago I did. Didn't kill me as the money was smaller then.
We all do it a different way and we trade way it makes sense to each of us. And no, I never noticed your stop disciplines. I simple don't trade like most do, I trade as per backtested results and some forward trading. Mixture of charting, TA and reversion to the mean with defined trend for stocks/ETFs and always hedged on entry and expectation of tops to protect open profits.
Last edited: