I have a couple of questions if you don't mind answering 1) How do smaller hft compete with the big guys like Virtu and Citadel? Is it possible to have a better signal than them or better inventory control since we already lose out in latency? 2) Were you always an algo trader? How did you get all the knowledge that made you so wildly profitable in 2007? 3) In HFT market making for delta one do you always end with flat inventory? 4) What are the most useful ML algos for trading now? And thank you so much for your time and everything you have shared in the past
While I respect your opinion this really goes against everything I've read and heard including my own experience. Most skills/field require far more than 6 months to become competent. Top expert level play requires years. You said yourself that trading is all or nothing - which I fully agree with. As such, I don't see anyone reaching such a level of proficiency in 6 months or even a basic level of competence. However, since you're recruiting graduates to your trading firm and probably have a training plan of sorts, we might be comparing apples or oranges. I'm talking about the lone retail trader who enters this business knowing nothing and with no guidance.
What I do personally trading now is more test ideas. Mostly automated. But making money isn’t the goal. Collecting information is. So I can’t imagine that helping much. I’ve never used a chart, indicators or ta. Either I look at market activity manually or at pcaps of raw data to dig for signals.
1. I wouldn’t try to compete on their primary trades. You can win by specializing. Also, there are many types of latencies to optimize. Single exchange, cross exchange, private vs public fills, easy logic fast, hard logic fast, etc. Most markets aren’t completely a winner take all and there are some internal constraints which make large firms sometimes do suboptimal things. 2) always is a strong word. But I was around traders very early in my career before I became one. Making money in 2007 did not take an incredible amount of skill. Things were pretty easy through about 2012. 3) some firms do or used to. I never have. I prefer less liquid products where that would be a problem. 4) I can’t be specific there as it’s a current huge area of focus for us. I’ll say that it’s important for each signal to make sense in isolation. There’s a lot of testing on signal quality and specificity before adding ML.
I understand the point you are making and I’ll reply intentionally trying to make the counterpoint. starting doing any business knowing nothing and with no guidance isn’t really a plan. For example, if I decide to ‘trade China’ I wouldn’t just show up, find a broker and start trading. There’s a knowledge acquisition period you need just to see if the business is viable. That can be long if you are slow rolling while doing other things or fast if suddenly it becomes urgent. After understanding 70ish percent of the problems then you can start. trading isn’t so much a physical skill like piano, violin, gymnastics which has inherent limitations on how fast you can push. It’s a mental skill which can be acquired rather quickly when the preconditions are all in place. I’d argue it’s on par with learning a new language. Say Russian. With 6 months of concerted effort and real intent you should be able to have a passable knowledge of Russian. If not, then language isn’t your thing and you should do something else.
How long would you say the knowledge acquisition period is, then? There's a major difference, though. If I want to learn Russian you pretty much have all resources to learn it at your hand. These days you can even learn it on your cell phone. Everyone is in agreement on what Russian is. For the amateur trader entering this business it's not quite the same - especially since most of us only have access to the internet. It's a jungle out there...
I is wouldn’t learn either trading or Russian through the internet. there’s way less misinformation if you 1) talk to people face to face or at least on phone 2) read a book. But not like a trading education book. Make sure the author made money from trading. Not book selling. strongly prefer 1) for trading. 2) for Russian. China knowledge acquisition is pretty long. I’d put it at 500-1000 hour. But not bad if you spend years slow rolling. Cultivating contacts and relationships over a long period of time.
Agreed totally. However, I'd say the main problem for most retail traders who don't enter this industry as professionals, but try to learn it on their own, is that it's hard to get that opportunity. For example, I'm based in Norway. Here, proprietary trading is virtually non-existent. Everything I know about trading is through the internet or books. Maybe you should add to your list: "Don't attempt to learn trading on your own or without a real time mentor." I agree totally on the opportunity cost. If I knew the seriousness/difficulty of the task and effort needed, I'm not sure I'd bother trying. Most, if not all, underestimate it.
I wouldn’t start a Hawaiian shirt business in Norway either... If forced to in that situation I’d try reaching out to people on linked in. To start conversations.