Backtesting Best Practise

Gaurav

Hi All, I'm very new to Algo world and using ctrader for about 6 months. I do have a good amount of time and computing resources at hand. I am well-versed with old-school RSI, MACD, S/R, EW, etc.

I tried a few bots recently and love the concept of backtesting my best strategies.

Wondering if you all experts and experienced Algo wizards can share a few best practices for backtesting as options and timeframes and backtest data can be overwhelming.

Again after doing all of the above granular control, things might change next week, month, bull cycle, season, war, or pandemic.

So what is the first starting point to find a clear winner and how should I drill deeper to make sure the winner is reasonably strong enough to withstand market moves?

1) Start with weekly or monthly backtest data and then drill deeper for 6 months or 1 year?
OR
2) Start with 1 year with minimal options and then fine-tune for weekly or monthly?
OR
Something else...!

Mike

this might help. https://ctrader.info/d/15-ctrader-cbot-backtesting

Gaurav

Thanks Mike for the above link, I have gone through all and realized that what I am referring to as backtest is actually what I meant for optimization to find the clear winner over RSI settings using the RSI bot or Moving Average using MA bot.

Hope someone can shed some light on how to best optimize these settings and timeframe.

Sebastian

This is standard for all bots. https://clickalgo.com/cbot-optimization

Gaurav

Thanks Sebastian.
I found this to be more useful to my concern: https://clickalgo.com/optimisation-guidelines

The exact thing I was looking for is here, as I was shooting all over the place earlier

Possible Scenario
One possible scenario of real strategy optimisation and one we recommend is to pick a date range in history for the optimisation run going back 4-months and ending 1-month before today's date, so that you will only be optimising the settings for 3-months leaving the last month un-optimised. Once you have the settings, you then run a backtest for the last month that was not optimised, this is fresh data and not curve or data-fitted in any way, if the results are good then you have a winner.