Thank you for your good comment. I don't really need to make changes or new implementations at this point as what I have is mostly working. Although it's a bit inefficient in terms of the implementation (relates mostly to speed and updating of data). I've always thought that at some point I would want to make changes to arrive at the optimal solution, but at the back of my mind I've always wondered how I could do that without giving away any secrets as I imagine a full re-write would require some explanation. This is particularly so as in a way I built the last part of the system on my own without any programmers involved. This opportunity came up mostly after a discussion on machine learning in some other forum and where I openly said that it's possible the best solution for me as a no-programmer is to collaborate with a programmer who have interest in learning the financial markets. I wasn't really actively shopping for it. Just thinking out loud. And then we started talking a bit in private. As far as I can tell this guy seems like an honest guy with good intentions and passion for such a project, but then again, people have disappointed me in the past.
I've demonstrated this system on multiple occasions on ET. Here's the latest one this Monday. I didn't think it to be related to this disussion, but seeing such negative interest in it I figured I could show the last one which worked out quite well. Most people who talk about probabilities don't actually back it with data, but just assumptions or "intution". When I make a forecast or prediction it's always backed by data. And that's what my model generates and gives me.
It seems like you didn’t read my entire first post on that chart because I specifically said 38-40 was a good sell coming up. That was said in real time, not after price had already dropped! And even then there was good downside potential. If I’m vague at times that’s deliberate because I don’t want to disclose all details about what I’m looking at, although I’ll say I’ve shared quite a bit too.
Maybe I’m missing the proposed exchange of value. My assumption is that your model hasn’t led to a level of profitable trading that you’re seeking. If that’s the case, it just seems to me you don’t have much to lose by revealing your model’s “secret sauce” in exchange for the programmer’s labor. Essentially, you are selling your model for the would-be cost of a programmer to make the changes you want. Of course my view is colored by my own experience. Unless the model comes with an ATS that is profitable, the model is not worth all that much without the knowledge and skill of the trader using the model. If my post seemed unsupportive, that was not the intent. I was merely expressing my jaded opinion that it sounds like a good deal for you. But I’m not in your shoes or in the programmer’s shoes for that matter. It’s great he feels he’s getting a lot from the collaboration. Enthusiasm is good.
Ok, fair enough. I deleted my post because in the end, I figured there is no point in me trying to convince you of something about your system when clearly only you know what the system is, but since you already read it and replied, I thank you for that. If the system is really good and you're killing it trading manually, then I imagine I would not give any parts of it away and just continue to trade it until I teach myself how to code it all up properly. I can see how a trader would get tired of waiting for signals and hence wants to go the automation route, but if its really that good, just one high conviction trade a day should lead to more profit than working a job. So I see no need to rush it if its really good and working well. The feedback from the one guy who said he was looking at the markets in a totally different way is encouraging, and even more reason to just shut your mouth and trade it manually until you can automate it.
One more question if I may. Yes, you said good sell coming up if momentum stalls, but use a stop. So the question is, how often will the system provide such a trade that could net 20 points profit with lets say a 10 point stop? I ask because of this. Below is a crop from my spreadsheet tracking my trades. This isn't the number of ES points I made, but rather what was possible for the trade that was entered. You see the different combos of 3 point stop or 10 point target, or 5 point stop and 5 point target, or 5 stop stop and 10 point target. The percentages are the win rates, and the totals on the bottom would be the total points made. Clearly, the most points made are from trades that are allowed to either hit a 5 point stop or 10 point target. This is based on 221 trades, so not highly profitable, but not horrible. A 39% win rate on a risk:reward of 1:2 isn't bad. So do you know what the stats for your system are? This is all I'm getting at with reference to my questions about the secret sauce. If you called out trades in real time, long or short, based on what your system says, what can I expect from taking the trades? This is what would tell me if I as a programmer want to copy your system and run with it, or if its something that a programmer would still need to put lots of work into figuring out how to use.
The whole thing you describe can be summarized in 1 word: EXPECTANCY Google it and calculate it. When I calculate it, I take at least 1,000 trades over a long period to have all kind of trends/cycles. I would not check for different stops and take profits. The reason why is: cut your losses and let your profits run. A good system should have a good stop system and the system should indicate what the market offers as profit and take the max out of it. If you put take profits you will surely sometimes miss huge profits because you got out to early. This misses can have a huge impact on your overall performance. A good system adapts to the strenght of the trend and tries to optimize the take profits in function of that trend. A take profit on 10-20 points can make you miss another 10-20 points because you got out to quickly. Fix take profits are static, while the market is dynamic. Your system should be dynamic for best performance.
Would you collaborate with a programmer if it meant sharing your system/research? -My answer is No unless I make him/her sign a confidentiality and non-competition agreement *and* have the time, energy and resources to monitor and enforce both of these agreements. The way I see it, your trading system/plan is proprietary and as proprietary as the secret recipe of KFC. Nobody should have access to that except maybe your children if you feel like it passing it onto them one day. You mentioned that your overall system is good and is still working and it's just few parameters that needs to be updated and some new features that preferably to be added. So why don't you just ask the programmer to code these specific features and mechanisms without involving the overall design of your trading system/plan? And then you just integrate them yourself into your overall system? C#, trust me, is not a difficult language to learn especially with the higher-level platform provided by the majority of the trading software companies, you can easily code these functionalities yourself when worst comes to worst. And when the absolute worst comes to the absolute worst, you can just leave the additional parameters and the features uncoded, imo if the overall trading system is still usable and you just to need to adjust your system irl when required.
Do not hire outside help if you have something decent. No NDA or something of this sort will help. Best to find someone who you know personally, a kid of your best friend. Still not great but better than nothing. Unless, even with code the person would be clueless. For example, if I would give you my algos code, you would be clueless how it works. If your code falls into this category than it might work.
Sorry but in my opinion (but I could be wrong), expectancy doesn't mean anything if you don't apply some parameters to it. What would is mean to say "is the trade profitable"? Does this mean it simply went 1 tick in your favor? How on earth do you capture that profit. What you suggest sounds good in theory, but there is simply no way to know that if 10 points is hit, the market will either continue, or drop shortly there after. I can skip taking 10 points and hoping for 20, but often times it will drop back down 5, and then to my entry, so I have nothing. Yes, there will be some major winners in there, but in order to get a 50 point trade, you will need to enter many times, and this is psychological not that easy to do for some traders, myself included. I am of course willing to be proven wrong though.