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)
I'll give two examples, maybe you don't do them with Market Replay (congrats if that is the case), but I know I have done them and probably everyone has too:
1. "The Do Over" - running through Market Replay over the same data more than once, in the name of "learning." Guess what, results always get better the second time, giving you a lot of false hope and unrealistic expectations on what your approach/strategy/system is capable of.
2. "Almost Fills" - convincing yourself that, even though your target was only touched or almost hit, you definitely would have had the foresight to exit before the trade turned on you.
Note these might not even be intentional, might not even be conscious decisions. But they sneak up on you and make the Market Replay results look much better than they would actually have been.
Indeed it is, and it has exactly the same issues, the difference being that the period of time being tested is fairly recent -- but you still don't know that the conditions that you had in that segment of the past will repeat in the future, when you're going to be trading it.
If you fit your strategy carefully to any particular span of time, including a very recent one, you are betting that nothing important will change. But the market changes all the time.
Bob.
------------------
Edit: I just saw that Kevin gave a better response than I did. But the issue of testing is generally the same: if you fit the past too well, you are going to set up for failure. So tuning your test to get a better backtest, or better replay results, whichever you do, can be very harmful.
When one door closes, another opens.
-- Cervantes, Don Quixote
Well, OK. The intent of the comments in this thread has been to be helpful, but if they are not needed, then no problem. I do hope you will be successful in your trading, and that you become one of the algo traders you speak of.
Good luck, and keep us informed.
Bob.
When one door closes, another opens.
-- Cervantes, Don Quixote
Will do. Yeah, all this is not easy to program. That was said in humor and not to get anyone touchy. I have really appreciated your inputs but you lost me now after becoming sensitive on something so silly. Also the only thing people on this thread keep expressing is why strategies or things cannot be successful. Good to hear negatives. But you guys gotta present positive things too. I'm sure you are making profits, I hope. It's bad attitude to say that if you dont agree with all that I say "in disguise of calling it helpful". Then I'm gonna take my ball and go home.
But good luck to you also my friend. Us retail gotta beat the hedge funds. I wish you the best.
It did sound touchy, didn't it? That's because it was, and I apologize for that. I should not have responded that way. So let me retract that sentiment.
There was a point, though. Yes, as you said, a lot of the commentary was how something wouldn't work, and it's hard to put those kinds of comments to use. But the fact is that backtesting is only going to do so much, and the weak point in it is that markets do change. So a too-exact fitting of a strategy to any period of time will often get you in trouble.
So what else do you do? Well, I'm a purely discretionary trader, and it's a little hard to translate from one world to the other. I can talk a little bit, for a while, about algorithmic topics, and then I have to stop, because the differences are too large.
The small point I had was to be cautious about any kind of becktested results, because of the curve-fitting possibility -- the fitting of a strategy to a set of circumstances that never happen again, even if it's very current and you're just using replay data from yesterday. The same issue, in a less-exact form, occurs for the purely discretionary trader too. Dealing with it is crucial to all traders.
This is not a plug, exactly, but many pure algorithmic traders on this forum have read and deeply benefited from Kevin Davey's book, "Building Winning Algorithmic Trading Systems," which you can get on Amazon. It's kind of pricey, but it is very commonly referred to by algorithmic traders here as a good source. If you want to dip your toe in first, he's got some more basic books up on Amazon also. My point in mentioning it is that it addresses the process of building an automated trading system and testing it. Disclosure: I'm going by the comments of others here. It's not my area and I haven't tried the book. However, the author is actually @kevinkdog, who is well-known here and who has left some comments in this thread already.
-----------------
There are some very accomplished traders on this site, who will chip in to try to help and who want others to be successful. It's perfectly OK to ask them to substantiate or clarify what they say, and it's OK to disagree too. But they often do have something to say.
Good luck, and I do mean this.
Bob.
When one door closes, another opens.
-- Cervantes, Don Quixote
It's all good man. I appreciate your refreshed sentiment. I agree with you about backtesting. Its no good. Market replay is better. That is what I use. Its the only best thing so far that's close to real time. It all has to be run to on real market like you said. Which I will do eventually. but I think its still good to test programs prior. I know you were trying to be helpful. So thanks. I have been trading on charts manually for over 10 years. Started this thread to understand automated strategies and programs and see what other people have achieved. Just curiosity. I ran market replay on 1 month. Now running it on 3 months. There is slight variation as you noted. nothing wild. Maybe I'll run it on 6 months. Takes about 25 minutes for 3 months of market replay. I use my fav indicators that have helped me for years manually. Gonna run it on real market soon. Programming eyes and human brain is hard but the positive side is human emotions are out the door which is awesome with algos.
I see similar statements a lot "algo trading mean no emotions."
A word of warning - this thinking is not true. Not even close to true.
PLENTY of emotions when algo trading, they just might appear in different ways, for different reasons.
Example: Your algo has lost on last 5 trades, should you let it keep trading, or should you turn it off? That can be an emotional decision, even if you feel the algo is still good.
The only way to avoid emotions in live trading is to demo/sim/paper trade - no real emotions in that. Alternatively, trade so small relative to your net worth that the money gain/lost does not matter (but then why even bother trading?)
If emotions are an issue for your discretionary trading, I can pretty much guarantee they will be an issue for your algo trading.
And believe it or not, emotions can also impact your backtest or market replay testing.
Don't engage in back-and-forth exchanges of personal remarks. Up to a point, every expression of opinion is fine, but when it becomes personal it is not fine.
You can disagree, but keep your disagreement within the forum's standards of civility and mutual respect.
I am deleting both of your recent posts. If anyone speaks disrespectfully to you, do not take it on yourself to answer them. Report the post to the mods by clicking on the dropdown from their username and we will address it.
And do not initiate an exchange. The internet is a free-for-all on some sites, but not here.
Please abide by forum standards of professionalism.
Thanks.
Bob.
When one door closes, another opens.
-- Cervantes, Don Quixote