BlueHorseshoe

Hi SIUYA, I wasn't aiming to suggest that the funds I mentioned were short term in their trading style (Winton may be in part, but I really have no idea), or imply that they offer any worthwhile pointers for those looking for daytrading robots. I was simply responding to what appeared to be an attack on the validity of mechanical trading by giving examples of high performing CTAs who have a proven audited track record that is about as beyond dispute as these things ever get (of course there are always the Madoffs out there!). People have been proclaiming the death of trend following for about twenty years now, but still those equity curves keep on pointing upwards so personally I doubt that their systems are broken. Thanks for what looks like a very interesting link.

If your statement is true for 'automated' trading systems (wherein a computer executes orders according to the system's rules) then it's true for mechanical trading systems in general. The post you made was interpreted by me to be an attack on the validity of mechanical trading. If you visit the Dunn Capital website, for instance, you will see that it clearly states that they operate 100% mechanical systems I quite frankly couldn't care whether Mr Dunn pulls the trigger personally, or whether he has a computer do it for him - the point stands that mechanical (and therefore automated) trading can be very profitable over the long term, and is not just a ruse of 'vendors', as your original post claims.

Ha, ha, ha! Ha, ha, ha! :rofl:

What utter rubbish! Go and look at the track record of a fund like Dunn, Winton, or John W Henry. Totally mechanical trading, and some of them have been doing it for over thirty years. Yes, there are probably a lot of crap, over-optimised trading systems out there, but the idea that trading systems as a whole don't work is complete nonsense.

Hi Onesmith, Are you saying that you would mimic the signal but using different algorithms? If so, then I misunderstood your original post. Nevertheless, I think if the code was sufficiently complex then it would be hard to mimic, even through some sort of parallel process . . .

What about if you tried for tops and bottoms, but you were willing to give the market plenty of room to move when you didn't get it quite right? In other words, if you entered positions when you anticpated a reversal, but didn't expect to buy or sell the low or high tick?

Adverse selection is what the books call it. Quite frankly, I was shocked the first time I tested a system and required the software to only fill orders when price traded through the limit rather than trade at it. Although I was fully aware of adverse selection, I had massively underestimated the effect it would have. There are ways in which the order queue can be 'jumped' (not literally), but these require a level of sophistication beyond anything I could achieve.

Hello, My current strategy would have handled the 2010 flash crash without any difficulties at all. Unfortunately I had only been trading for about two weeks at that time, and was busy losing money trying to daytrade using hopeless tactics from the John Carter book. What's more, because the day of the flash crash coincided with the day of the general election here in the UK, it got absolutely no news coverage whatsoever! However, I do understand the point that you're getting at, so fast forward fourteen months or so and you have the huge sell-off off last August. My strategy (which I was trading in real time) handled this very poorly indeed. So if you go and look at JSwanson's original equity curve for buying pullbacks, and see the massive drawdown that occurred, this is roughly what happened in my real trading account at this time. Subsequent winning trades have since virtually erased this loss, but a similar event may well occur tomorrow. I have spent a hell of a lot of time since last August looking at possible strategy revisions, including all manner of stop losses, but have not found anything that is satisfactory or robust. My conclusion: because I trade only with money that I can afford to lose, I am happy to risk trading without a stop in order to swing for the bigger gains. I hope that's a helpful answer.

I agree with your post, but . . . Here is a system with a 100% win rate. I'm going to share it with you completely free of charge or obligation. I have not made any money from this system for myself, but please feel free to use it in your own trading. Should you decide to use it, make sure that you follow the system rules exactly. 1. Buy the DJIA in 1903. 2. Sell your position tomorrow on the open. 3. Don't take any more trades. I hope that's helpful and will lead some of you further down the road to riches. Oh, and if you experience any difficulties implementing this strategy, please DON'T pm me.

Except that this is financial market data, so that's not strictly true. Otherwise Bollinger Bands would work a lot better than they do.

Hi Tradewinds, Without going into the DLL process, there are a few things you can do within TS to make the indicator more secure. A determined hacker will still get what they want in the end, but they'll need to be pretty damn committed about. Break the indicator's code down into as many modules or 'functions' as possible, so that the indicator itself becomes an empty vessel. You can now protect each function individually. Use passwords that employ the maximum number of characters TS allows. Use a combination of letters, digits, and symbols. Don't use a word that is recognisable in any language. Doing this will mean that, even running pretty fast software, the code will take about three days to crack. And they'll have to do this with each individual function in turn, which could take weeks, or even months, depending on the complexity of the indicator. Take steps to make the indicator more complex than it needs to be, and introduce empty functions near the start of the code. As most hackers start to crack these, they will quickly become pissed off and fed up. As I said, this makes things harder to crack, but not impossible.

Hi Onesmith, I'm not convinced by this. Can we test you?

So who sells options, why do they do it, and how (if at all) do they go about limiting risk?

Yes, you need to seperate 'buying the low tick and selling the high' from 'successfully trading reversals'. The former would be the ideal scenario. Although it seldom resembles the ideal scenario, the latter can still be a profitable reality, and is not just a delusion of vendors. Anyhow, there are plenty of vendors around selling other ideas (One of today's highlighted threads starts with a quote from a Covel book on trend following for example).

I am aware that I appear to have hijacked JSwanson's thread with digressions - so no more posts from me here now.

I can't help you with an explanation, I'm afraid. What I do know is that if this is the easiest arb on earth, you've just shared it with several thousand other traders . . . Thanks! Bluehorseshoe.

Hello, You would calculate the win/loss ratio exactly the same as you do for any other system. A win occurs when a position is exited for a profit, and a loss when it is exited for a . . . loss. Just to be clear - I am not saying that I will allow a position to run indefinitely against me. Hence I still have losing trades when I exit at a loss. I am saying that I don't have a hard stop in the market. This means that the theoretical size of my loss, if the market moves hard against me, is the entire value of the contract for longs, and infinite for shorts. I am willing to bet that though this is the case in theory, it won't be the case in practice. With regard to how deep a trader's (or fund's) pockets are, please see my response to MightyMouse above - you don't need deep pockets to do this, you just need to trade within your means.

Hi Joshdance, Reading back through your post it seems that I have probably misunderstood you and mis-represented what you were saying. Sorry about that! I think that my statement relating to volume distributions is perhaps also unclear. By no means was I trying to imply that anything like a textbook bell-curve occurs with any regularity. I was suggesting that volume typically diminishes as the point of support or resistance is approached. There is a (slightly perverse) sense in which this is definitionally true: there is always less volume one tick below the low tick than there is at it. I am not claiming that this really occurs according to any ideal model. The 'bell curve' may well end up looking like a whale's silouette, but even a whale has a nose! Rationally, I cannot find any reasons why a market trading at ask would cease rising other than because no buyer will consent to buy at one tick above the high, or because sellers start offering below it; in either case, there is insufficient liquidity at higher prices, and the market will trade towards where there is liquidity. "Sure, there is a low tick and a high tick, and unlikely that that high or low has lots of volume transacted. But typically there will be an area around where there is high volume." The point I am making is that the area won't just be "around" - it will always be above the low or below the high, and the market will have traded back towards that liquidity leaving the high or low tick as the tidemark we call support or resistance. The support or resistance occurs because the market has encountered a lack of liquidity, and has headed back the way it came to areas of higher liquidity.

This could be a really interesting thread to follow. My understanding, quite different to Joshdance's above, is that a market gravitates towards liquidity pools, and not away from them. So I expect areas of support and resistance to occur where there is little interest (or widespread agreement) from buyers and sellers. From a historical perspective, I would expect support or resistance to be accompanied by low volume. The absolute S/R point (one tick above the high/low) has zero volume of executed orders. If you look at volume within a trading day, the above is exemplified by the gaussian distribution that MP traders pay so much attention to - the closer you move towards the high or low of the day, the less volume is typically executed at those levels. Ironically, of course, a lot of volume is often executed at the prior day's high/low. However this is all just my understanding (and doesn't actually guide my trading in any concrete way), and I have had price action traders who spend a lot of time watching the DOM tell me that my understanding is misguided in other threads - so my word clearly isn't gospel! Hope that at least provokes further discussion on this topic. Bluehorseshoe

Hello, Equity-curve trading is an interesting topic that remains fairly esoteric - it doesn't seem to be mentioned much on here. One very simple approach, rather than an 'x' drawdown in fixed dollar terms, is to cease trading when the equity curve crosses below a moving average. Obviously this MA could also be optimised, although I am not aware of any software for doding so - maybe some of the more sophisticated portfolio analysis programs may be capable of this? Here's another point you might like to consider: a strategy that is profitable will have an equity curve that is mean reverting. This means, in theory, that you would want to increase trading following a drawdown, as you can shortly expect a renewal of the upward equity curve. This is only true in so far as you can be certain that your system will continue to mantain a positive expectancy. If, on the other hand, you had doubts about this, then an 'x drawdown' type rule for cessation of trading can provide an objective way to define 'system failure'. I hope that's of some help to you - good luck researching a complex and seldom discussed topic! Bluehorseshoe

Thanks Nemesis - you've described in one word what took me several paragraphs to explain! I wouldn't say that I'm under-leveraged against my account though. I would say that I'm underleveraged against my 'net worth', if that makes sense? The obvious downside of this is that I'm not making anywhere near as much money in real dollar terms as I would be if I took advantage of the opportunities to use greater leverage. The reason I avoid this is because although I am prepared to weather what many would probably consider a very significant risk in percentage terms (a risk that any single trade could potentially wipe out my entire account), in real dollar terms I am not prepared to risk my net worth in this same fashion. Bluehorseshoe

Ha! And then there's that category who fleece sheep but don't have any trading losses to cover because they don't actually trade . . . OneSmith has provided me with valuable coding assistance recently and didn't attempt to sell me anything, so I'll happily second his statement above and recommend seeking help from him with coding.

I suppose what you really want is a stop-loss on your equity curve - a rule that states to cease trading if a certain drawdown is reached (including intra-trade). Just a thought though, and equity curve trading is a whole other topic . . .

Hi MightyMouse, Thanks for your reply. I totally agree with you about inference from historical equity curves - there is absolutely no way that you can know that the equity curve will prevail. But if the trading concept here (buying three day pullbacks) ceases to work, then it will cease to work both with stops and without stops. As it currently performs far better without stops, then it is not an unreasonable assumption that its failing performance would be better without stops. The equity curve is a function of this strategy (which doesn't contain stops). So I don't believe that it's simply a case of adding stops to the strategy; you'd be creating a whole new strategy (with stops) that just happened to share the same entry criteria. I personally trade without stops. If the S&Ps fell to zero today I would lose most of the money in my trading account - this wouldn't be a huge delight, but it wouldn't be anything more than a dissappointment - I can afford to lose that money, otherwise it wouldn't be in the trading account. So there are other ways to control risk beside using a stop loss that massively reduces the perfomance of your strategy. Not that I'm desperately trying to convert anyone reading to stop using stop-losses - that would just be irresponsible. In an effort to try and provide a less confrontational response to your original question, I'll re-run JSwanson's strategy and post a report for different stop-losses within the next few days. Hope that's helpful. Bluehorseshoe.