History tends to repeat itself. I am not experienced enough to debate this topic !
Ask @Vonasi. He might tell you his opinion.
Price Action history is like clothes fashion … repeats, but never quite the same?
@Nicolas some indication of optimisation is within the algo itself.
// **** 4 steps for success ****
// ** STEP 1 – Strategy idea and alignment
//Defining an edge or an idea, and enunciating the strategy including settings e.g. timeframe, index, stoploss
// ** STEP 2 – Source code optimization
// IS/OOS-optimization of the source code parameters using 60-70% IS and 40-30% OOS.
// ** STEP 3 – Walk forward testing
// 60/40 WF-test over 5 iterations. All periods must be over 50% WF-efficiency. Total efficiency over 80%.
// ** STEP 4 – Demo, trial & Live
// Algos tested in demo for X trades or 10k bars before going to live, also for X trades or 10k bars
Thank you for this info, so it matches the idea that I had indeed. The development looks quite decent and sane to me. That's the way most of them here work for their strategy development. So indeed, if we use the latest version with the latest parameters on a backtest on 1M candlesticks, then it seems normal that the performance in this OOS data is less good than on those we validated with the last OOS occurrence. This being the case, it will of course be necessary to return a new version to the period defined by its iterations which validated these OOS, this is what the different iterations are for. So as it stands, I don't know how many months that could represent.
Summary of all the algos backtest over 1 M bars – I believe these are correct but don’t guarantee it.
Ask @Vonasi. He might tell you his opinion.
In my opinion (for the little that it is worth) every strategy is curve fitted to historical data because that is all we can ever do. Will history repeat itself – yes sometimes and no not at other times!
History is relevant. The candlesticks created before the invention of the computer are very different to those after its invention. The candlesticks after Trumps election are very different to those before and so on. All indicators are created from price movements or comparisons of price so comparing today’s indicator values or movements directly to those in 1972 is pretty pointless.
A strategy can be a winner in 1972 and a winner today but we must try to fully understand why it is considering that we know that historically price movements have changed drastically over that amount of time. In fact it was easier to make money before the invention of the computer but we only know that now because the computer has been invented! If we were drawing candlesticks with a pencil and calculating an average with an abacus then making money in 1972 would have been much harder than we think it is today.
So don’t rely on history but as it is all we have to use it is the only thing we have to rely on!
Summary of all the algos backtest over 1 M bars – I believe these are correct but don’t guarantee it.
Can you post the attachment once again, please ? I guess it got lost somewhere…
With attachment as it seemed it did not work first time.
I agree all we can do is to find a strategy that fits the curve as closely as possible, and hoping it will continue. Going back to 2010 is not really going back when computer did not exist 🙂 and I assume that in 2010 already most of the trading (including professional trading) was already controlled by algos. I also assume this is a constant game of police and thieves, one finds an edge in response to an algo trading and it is successful for a short while, until it gets found out …
So in that respect, recent events have more weigths than past events. for a short while. We can also hope that a certain logic can be applied, and that this logic remains true.
Only one thing is certain is that there is no logic in the market !!! 🙂
I also assume this is a constant game of police and thieves
It is a constant game of buyers and sellers. For every buyer there has to be a seller or lots of smaller sellers if it is a big buyer. This has never changed but the numbers have just got bigger both in price if it is an index and bigger in quantity of buyers and sellers for everything that can be bought or sold. A look at any commitment of traders chart will show us how the small guys are doing what they think is right and the big guys are just using their massive equity to hoover up and make a profit off the small guys no matter what direction price is going in.
It would be so much trustworthy if that could be a part of the selling package.
There is no trust and reliability in trading. I think this ought never to be forgotten.
Yes excel. Now with totals
Any reason why you didn’t include the ‘PRTA’ Algos in your excellent Table of results?
It is included, the first row. But sorry I did not keep the exact naming convention , first 10 are the old PRA algos and the next 8 are the new ones PRA+ (I did not do Nikkei and it seems I missed another)
You should be able to easily match.
Hi Guys,
No more answers from @ProRealAlgos ?
We are here. We just don’t want to interfere with the discussions but let people make up their own mind 🙂
