Welcome to NexusFi: the best trading community on the planet, with over 150,000 members Sign Up Now for Free
Genuine reviews from real traders, not fake reviews from stealth vendors
Quality education from leading professional traders
We are a friendly, helpful, and positive community
We do not tolerate rude behavior, trolling, or vendors advertising in posts
We are here to help, just let us know what you need
You'll need to register in order to view the content of the threads and start contributing to our community. It's free for basic access, or support us by becoming an Elite Member -- see if you qualify for a discount below.
-- Big Mike, Site Administrator
(If you already have an account, login at the top of the page)
Trading: Primarily Energy but also a little Equities, Fixed Income, Metals and Crypto.
Frequency: Many times daily
Duration: Never
Posts: 5,048 since Dec 2013
Thanks Given: 4,384
Thanks Received: 10,205
If you pick the right systems it's profitable, but if you pick the wrong one's it's obviously not. Obviously if it was easy everybody would be doing it. I think I have shown in this thread that picking winning systems is difficult! You do make a great point though, that there doesn't seem to be anybody claiming to be doing well using this. I have several people send me direct messages about isystems, and I haven't got the impression any of them have consistently done well either.
Right from the beginning I stated it was my goal to discuss how to select the systems rather than specifically talk about the systems I picked. I obviously didn't do this so that I could then fake my results as I've made it clear I actually lost money. I must admit when I started I was expecting that this would be a money making not losing adventure, and didn't want to be responsible for people copying me. I also didn't want people to think I was favoring specific systems or to be overly criticising systems. As you will have read most of my comments talk about performance of systems as a whole or in groups, rather than as individuals.
Looking at your systems, all four have almost perfect looking backtests. This makes me very concerned about over fitting and one of the reasons I value the "since tracked" or "out of backtest" data so highly.
MiniSp Sevilla StopTakeProfit Intraday is very new, and already arguably not performing like its backtest.
Mandala Gasoline is performing exactly in line with it's backtest. I find this interesting in that Gasoline/RBOB is one of the most volatile futures contract out there but this system has small wins and doesn't experience any wild swings.
Swing 6 _ 30y T-Bond ZB again is very new with only 5 months of real data. Does it concern you that in the 27 months of backtest the only losing months were <$385> and <$432> but in the first month of non backtest data it lost <$1269>?
CIRUS ST STOP 1% NASDAQ ODOR is another (rare) system that seems to be performing very much in line with its backtest.
I bought into this isystem last month and promptly lost on 3 trades. Their performance info made it look like all I had to do is sit back and wait for the money to roll in. ha ha. Now I'm wondering if I walked into a buzz saw??
This looks very similair to what I was saying to @InvestorZ.
Here's your system. As you can see unless June is at our above average performance it will already be out sample after just 6 months. Remember the darker grey band shows a one standard deviation performance band, and the light grey two standard deviations. Hence if the backtest is predictive of future performance we would expect 95% of system performance to fall in the light grey band.
Trading: Primarily Energy but also a little Equities, Fixed Income, Metals and Crypto.
Frequency: Many times daily
Duration: Never
Posts: 5,048 since Dec 2013
Thanks Given: 4,384
Thanks Received: 10,205
In the last several posts I've talked a lot about "out of backtest" performance versus in sample. Obviously this is also highly related to my initial (failed) hypothesis that systems that perform similarly in OOS vs IS are what we want to trade. I have lots of idea's on where I could take this, but last night suddenly thought why not create a variable that measures performance in relationship to the backtest. Simply
(Actual Performance - Expected Performance)/(Performance Standard Deviation)
So anything less than -2 would mean we are more than 2 SDs away from expected performance. I suspect that this as a standalone feature will be non-predictive but hope that when combined with others it could be valuable.
As stated previously I agree. Unfortunately the only data I have for the inactive systems is the 'Out of Backtest" performance that I have previously saved. I do not have a record of their "in backtest" performance and have no way to get it now. So if I perform a calculation like this, I can only ever use currently active systems.
The actual calculation involves what I thought would be a simple code modification. Unfortunately I'm getting lots of unsuspected html errors. I don't know whether this is a code issue or an internet scrapping/isystems website issue. Time will tell.
been tracking Google Trends Monthly NQ and Google Trends Weekly NQ since Dec 2018, vastly different results even on the same logic but different parameters, if such structured systems by the professionals do not work in the long term, what is the chance of the retail that tries to build these on his own?
Trading: Primarily Energy but also a little Equities, Fixed Income, Metals and Crypto.
Frequency: Many times daily
Duration: Never
Posts: 5,048 since Dec 2013
Thanks Given: 4,384
Thanks Received: 10,205
One of the drawbacks of the way isystems work is that it's impossible to use a walkforward development. While not everybody uses walkforward I do believe it's more common than not with successful systematic traders. That could be a factor as to why so many of these systems appear to struggle.
Looking at those two systems specifically...
The weekly after only slightly underperforming for the first year of tracked, has really crashed in the last 5 months.
Conversley the monthly has been over performing for most of it's 2+ years of tracked performance. Unfortunately in the last seven months it's experienced two large drawdowns never before experienced. Saying that, it made the money back in both cases and is currently at new highs and has made more per month since being tracked than it did in its backtest.
Trading: Primarily Energy but also a little Equities, Fixed Income, Metals and Crypto.
Frequency: Many times daily
Duration: Never
Posts: 5,048 since Dec 2013
Thanks Given: 4,384
Thanks Received: 10,205
Should add, none of those charts include June, and both systems are up in June pretty big.
Also all the Google systems are by the developer Investmood. I searched around and found their website investmood.com which is in spanish (I think) but that translate button in chrome comes in very handy! Reading their blog I found several interesting passages
In March of 2019 the Big Data System of Algorithmic Trading " Google Trends Monthly ES " has made two years verifying its predictions in the real market. This system takes long or short monthly positions based on an IA model based on what the population of the entire world looks for on the Internet, therefore based exclusively on investor sentiment. We observe that the ROI over these two years has been 52%, and that of the 24 months in which it has been operating only in 4 it has experienced losses (83% success rate of the model). This is the InvestMood system that has a better Trading Motion rating, showing that the investor sentiment measured globally is a valid predictor of the evolution of the S & P 500
and
We welcome the new big data algorithmic trading system developed by InvestMood and that is already available through Trading Motion and iBroker . In this case it is the Google Trends Weekly RTYwhich differs from the rest of InvestMood's systems in that the predictions do not come from a single model of artificial intelligence, in this case the predictions about the trend (rise / fall) of the Russell index, over the next week, result from the combination of 6 models of artificial intelligence on the weekly and monthly trend of the main American indices (Dow Jones, Nasdaq and S & P500). We have used the predictions made by these 6 models in 2018 so there is no backtesting and what the historical shows is simply the result of the combination of these models throughout 2018. It also has the following characteristics:
Name: Google Trends Weekly RTY
Predictor: Google Trends
Stop Loss: 6%
Take Profit: 9%
Algorithm: Bayesian Networks
Periodicity: Weekly Training: C omombination of models Instrument with which it operates: Future Mini-Russell CME (USD) Probability of entry : Swing
Note: Trading Motion is the European equivalent of iSystems
Interesting background, at least we know that there is some usable application of AI and sentiment in trading, if this really sustains, not sure if the AI incorporates machine learning and adapts along the way. Ideally it should work longer than traditional based technicals and therefore the Weekly_NQ should recover back to new equity highs soon
Trading: Primarily Energy but also a little Equities, Fixed Income, Metals and Crypto.
Frequency: Many times daily
Duration: Never
Posts: 5,048 since Dec 2013
Thanks Given: 4,384
Thanks Received: 10,205
Finally something predictive?
So this is the distribution of the 22597 samples of twelve months of tracked data versus the backtest, for systems that are still active.
Only approximately 9.5% of systems have performance better than expected from the backtest!
Median performance is -0.46 meaning 0.46 standard deviations below backtest.
And here's what our results would have been if each month we traded the system that had the highest/lowest/nearest performance in the previous 12 months when compared to the backtest.
