I don't do walkforward testing. I know there are people who think I should be burned at the stake for saying that, but my reasons are: 1) I *never* have has much data as I would like anyway, and 2) I am trading my own money and don't have to convince anybody but myself. That said, I think I pretty well understand most of the problems of curve fitting, survivorship bias, lookahead bias, hidden risk, etc., etc. I use fairly simple simulations to decide whether or not to put real money at risk. I do almost all my own programming, and I prefer to keep things simple -- I can be a bit dense at times. I also don't always test my software as well as I should. I guess that's what my employers and clients were complaining about all those years. The more software I write, the more that can go wrong. I'm not sure there is a specific rule that determines when a trading idea has been proven. For my own money, I go by the seat-of-the-pants. But I have seen investors throw good money at some pretty half-baked ideas too that were pitched by some smooth-talking salesman. Just some random thoughts to keep the discussion flowing. . . Good thread!
OK, this thread is interesting because I'm 99% sure it's not bullshit. So I figure I'll start my ET posting career here. I guessed the reason more or less correctly (I guessed it was index funds rather than funds with a S&P-only charter, but close enough). So now I want to back test this. I can easily enough get dates off the S&P website, get daily bars for (almost all?) the symbols in question off of yahoo or google, and start figuring out if my stock trades executed or not. But I see some problems before I even get to deep statistical and confidence issues: 1) if I entered the limit order, and the stock opened the next day at a price where I will execute, what exactly happens? I know what my worst case price is, but I don't know what price I really get. I guess if it works with worst-case, then that's OK. 2) I don't know exactly what time my order executes some times, so I don't know what prices I get on my hedges if I need them. 3) If I have two opposing orders (the normal case) and one executes and the other doesn't, do I put on the hedge until the other does? Or is it OK to be net short or net long for some period? That's not really clear from the system. Now, given that this effect is semi-well-known, I have two questions. One is, if you're a fund manager why isn't there a rush to make your fund's move first on open the next day, thereby compressing the price movement to the point where it's untradable? And two, why does the stock wait until after the actual delisting to rebound? Shouldn't that happen as soon as the majority of fund managers get around to making their trades?
Thanks! I use MS Excel to see how my trades are compare to random outcome, and I think that is quite helpful. So, what other analysis could we do with Excel? PA
Keep in mind that I don't actually know that's the explanation... that's my best guess. If it's correct the reason could be that the finds have to 1) have the change done by the effective date and 2) minimize price impact and 3) move even more money than you might imagine. the end result of that might be that they have to use the whole period of time to do the trade. Let me stress... I don't know that for a fact... but it's a reasonable guess. As for your other questions, they are great ones. 1) I guess assume worst case or take the opening print. be consistent. 2) get it close. be consistent. being exact shouldn't matter. 3) well... here's where it gets a little more messy. we dont trade this exactly like the rules i laid out... we do some legs with options (basically the some shorts because i'm willing to eat the theta because i'm a little worried about a short popping 400% against us. if that happened and you were short 100 deltas two things would help you: 1) gamma would help you out and 2) implied vol would almost certainly spike which also would help.) net net... the options BS has cost us money. we also leg into these things... like buying some above the price and messing around with the hedges too (because i'm a short term trader and think i can get a better price. waste of time net-net.) this is one reason why me posting my results will not help this discussion. basic answer to question #3 would be that you should pretty much never be unhedged. it's not about market exposure understand, but the system was backtested on EXCESS returns.
Making consistent sizable returns daytrading S&P futures is a lot harder than you think... in fact it might be the hardest game in town... but for some reason it's what 90% of internet forum readers want to do. maybe it's partially about being a badass ("I'm an S&P trader")... but it's much harder than trading has to be. I guess I think of markets as being partially random and partially "signal" for lack of a better word. The S&P futures are usually one of the most random and noisy markets out there. I can trade them and make money, but doing all of this other stuff is more profitable... and you learn a lot more which is also part of it for me i think. I think the internet public has been sold a bill of goods about making a living daytrading ES... it's not a great idea in my experience.
A safer bet might be to use an ETF which is in the same industry as the stock you are hedging. For example if the stock was a financial stock you might use IYF or IYG as the offsetting position rather than SPY (IYF is a Financial Sector ETF and IYG is Financial Services). Great thread btw.
thank you. but it's not about hedging... it's about excess returns... so you have to offset with the market.
Sorry about not clarifying that... it's a term from academic literature mostly that means "returns in excess of (over) the market". For instance, if I buy a stock that goes up 5% and the S&P goes up 3%, that stock had a 2% excess return. Note that a stock can even go down and have a positive excess return... in which case you DO make money if you have both legs of the trade on. It's just another way to talk about a spread really. Does that make sense?