Hello, I have been visiting this community for quite some time, and even asked several questions, for which I was getting interesting and useful answers, so I am back with another question, but this one is a bit long-winded, even I tried to make it as a tl;dr.
First, I understand that this industry appreciates concise and short statements, but I think that for this question, the context matters.
Second, a brief introduction: a Ph.D. in Theoretical Physics, experience with working on quite interesting projects, active DoD TS clearance. Translation: often times I know what I am doing and saying ).
The context: Some time ago I decided to see if my knowledge of stochastic and nonlinear systems could be applied to describe how the markets work at the “atomic level”, in a sense of the following chain of events: 1)the Order Flow interacts with orders at the Order Book, 2)these interactions start at the Top-of-the-Book and “propagate” to the adjacent levels, 3)these levels deplete/repopulated, 4)the instrument does (or does not) move, 5)the Order Flow may (or may not) change, 6)process repeats in a similar, or maybe in a different fashion.
If trying to devise a solution for such complex problem sounds “a bit strange”, you may compare this task to mastering chess, or GO, and it will start looking just fine.
Long story short, after a noticeable time, mistakes, errors and a decent set of knowledge acquired (market mictostructure, regNMS, HFT, Gamma, 0DTE, the lot), we (me and another Ph.D.) developed an automated framework that in real time deciphers and analyzes the processes at the tick level. Obviously, the system does this analysis at a very detailed and complex level, but the outcome is a prediction for the most probable next move of the instrument, which may be anywhere from several seconds to tens of seconds.
Most important for us is the fact that by deconstructing and quantifying the processes at tick level, we can tune the strategies we built if market regime changes. The reason, is that the strategies are built from our standardized modules, and more importantly we understand what processes each module describes and how it describes it. The analogy for this approach might be a comparison of how DNA is built of only 4 nucleobases (although our “vocabulary is larger, but still less than 15 important modules).
Obviously, we made mistakes and took several wrong turns, some of them were probably avoidable, had we known “what we did not know that we did not know”. We also got several good advices from different people, some of them from “anonymous” accounts here at WSO, for which we are extremely grateful.
The question: Here is my question: given that “occasional advices” cannot substitute a proper guidance, how unusual for the financial industry is to ask for one? We are not asking for funding, or something along this way. I think we have what we need, at least for now. What I would greatly appreciate is akin to a “scientific advisor”, who does not solve the problems for you, but based on his experience, he may advise which route is worth trying, and which is not. Thank you. Cheers!
Based on the most insightful WSO threads and discussions, your question about seeking guidance in the financial industry, particularly for a highly specialized and technical project like yours, is not unusual at all. In fact, many professionals in finance, especially those venturing into quantitative finance, algorithmic trading, or market microstructure analysis, often seek mentorship or advisory relationships to navigate the complexities of the field.
Here are some key points to consider:
1. Advisory Relationships Are Common in Finance
2. How to Find the Right Advisor
3. What to Look for in an Advisor
4. How to Approach Potential Advisors
5. Alternative Routes for Guidance
6. Final Thoughts
Good luck with your project! It sounds like you’re on the brink of something exciting, and with the right guidance, you could make significant strides. Cheers!
Sources: Q&A: Buy-Side Analyst, Advice for Equity Research Associates - Part I, Strategic finance?, Biotech finance part 2: valuation methodologies and modeling considerations, Are you guys happy with life? Help me please.
Unfortunately this isn’t a novel idea but rather a quite mature approach. Your credentials (PHDs / clearance etc) are only a positive indicator if they come alongside a track record. What is your realized sharpe, scalability, t-cost, market impact etc. + saying you’ll anticipate regime changes is… concerning hubris.
Re seeking advisors - well if you have a real & novel method with somewhat credible quantified results - it should be no issue to raise money given credentials check out. Frankly - the first thought is - you probably don’t. But it’s always good to seek advice & want to learn. IMO (particularly with quant stuff) you won’t find many willing to share specific methods & strategy suggestions unless you are working for/with them - scale is a constant constraint & sharing niche strategies tends to evaporate the alpha oppty
I’d seek roles at a market maker or quant fund or raise a little friends/family capital & put your savings to work building a track record if you know it works
Re seeking advice - id try college alumni first & any friends of friends of friends.
Thank you, Anchor. Appreciate your comment. May I ask you to elaborate more on why, "on the first thought", you think that I do not have a novel approach. Just wondering, as the problem does not appear trivial, at least to me. If the idea is not novel, and the methods of solving this problem are mature, so nothing new is expected, then, could you please suggest where can I read more about it? Also, I never said that my system "anticipates regime changes". I would never say that, as this is a proverbial "million dollar question". Let me paraphrase what I said. In case when the predicted outcome starts diverging from the real instrument moves, then, based on the detected "signal signature(s)", I know what module(s) need to be deactivated, and which need to (re)activated to get system back on track. No hubris here, just a "predictor-corrector" approach. The only difference is that I do know what branches of the decision tree to cut/replace. Thank you. Cheers!
You will, quite frankly, struggle to find resources in depth as I would imagine you would like. There are, generally, two ways of making money at the faster end of trading, which is what yours seems to be - be quicker than everyone on the easy signals (ie snipe the obvious good trades whilst they are unfilled), or be more accurate than everyone else (ie identify the good trades before they are obvious). At a timeframe of seconds-10s of seconds, you are firmly in the accuracy camp. The problem is signals here became significantly more valuable as they are your only moat, so you will never find up to date information on how exactly anything is done. Any advisor you could find that still has skin in the game will have an incentive to steal your work (pretty obvious as you'd decay immediately), and I question whether anyone who doesn't has the knowledge to accurately guide you or give you a better view of the current landscape. As many others have said, your best bet here is to run it live and see how it performs, which gives you a leg to stand on if you ever wanted to take it to a firm to run it, ideally a collaborative shop rather than a pod if you still valued guidance.
What do you consider to be "credible quantified results"?
Is this question to me, or to Anchor, as I did not use this particular term? Can you please clarify? Thank you.
No, it is directed to Anchor
Can you tell me if a stock will beat the bar or miss the bar?
Thank you for a good question. I did not think about this task, as I tested my platform mostly on the futures. Let me think about it. What I certainly can do, it to tell how the buying/selling patterns are changing the internal dynamics of the Order Book levels. In other words, if there are some systematic strategies that try to obfuscate their activities. This problem looks like an interesting one, but I would need more context in terms of required time horizon, how liquid the stock is, etc.
Sounds like a “no.” Alpha comes from betting on bar beat or bar miss.
These alpha models are only useful if part of a complete strategy that has been running live for at least a few months. If your predictions are a few seconds ahead then this is more of a prop strategy than a MM one. As said earlier, people will be interested in the strategy's performance and not whether the models are novel or not (often the profitable ones are not).
I had a similar background as you once a long time ago and can discuss more over PM.
Thank you for your response. I will be interested to continue this conversation over PM, which I will be sending in a minute. For now, as the general comment, I would say that this is exactly we are planning to start in the next week or so. Thus, the questions that I have are pretty much along the lines you outlined: what should we use to evaluate the performance, what instruments would be better to start with, etc. We had a bit of experience with ES and NQ, but I do not think that such highly leveraged instruments would be a good starting "portfolio". Thus, I am more inclined to go into ETFs, and probably will start with SPYs and then expand. Anyway, let me PM you, so we can continue on a more detailed level. Thank you.
Good afternoon Mr. Simons, hope all is well
Did that and saw GPT either deviating from the topic I wanted to cover, or hallucinating .. Both are not good, as I wanted to be precise. This is why I said from the beginning - tried to reduce to tl;dr, but sometimes the context takes over the brevity,
I find people often overestimate GPT’s ability to deal with complex tasks
Consequuntur reprehenderit ut doloribus aperiam rerum eum. Ducimus earum minus earum id. Quibusdam quia fuga vitae nihil. Autem velit tenetur odio temporibus. Blanditiis sit dolorem minus facilis voluptatem aut.
Molestiae ut nostrum illum quis assumenda aut. Facere ut necessitatibus consequatur modi. Sit quis accusamus et ut eius laudantium. Tempore esse quae impedit ullam. Perspiciatis illum est quia sapiente sed magnam.
Doloribus animi ea non at quasi. Et unde itaque vel aut. Deleniti vitae dolor nemo quae dolores. Tempore distinctio qui eos mollitia nemo. Distinctio itaque consequatur laboriosam facere id autem ducimus. Quae qui enim voluptates aperiam praesentium iusto dicta.
See All Comments - 100% Free
WSO depends on everyone being able to pitch in when they know something. Unlock with your email and get bonus: 6 financial modeling lessons free ($199 value)
or Unlock with your social account...