all-prop

@quant_arb vol normalized log returns?

@Solving_Inv For how I set up my system: 1/5/15min forward return target alpha control the speed of the TWAP (usually around 1h TWAP) Haven’t added sub 1 min alphas to market make around yet but I’ve built them before when I ran HFT strategies on perpetuals so it’s on the TODO

A large part of the answer to “what is an execution algorithm” is not stealthy iceberg orders and sneaky flow hiding, it’s just having great short term alphas you use to inform whether to execute faster or slower.

@quant_arb any logic where you might warehouse more risk than the TWAP would suggest bc of <1min/5/15 alphas?

@Mang0_trad3r @macrocephalopod @BobEUnlimited @nishantkumar07 i think. With same reasoning a sharpe 3 strats should still have losing years 2.5% of the time

@Mang0_trad3r @macrocephalopod @BobEUnlimited @nishantkumar07 assuming normal distr. yearly returns. Looking for P(R>0), which is P(R - mu / sigma > 0 - mu / sigma) which is P(Z > -sharpe) which is P(Z < sharpe) which is P(Z < 1) where Z is standard normal variable which is (look it up) ~0.83 which is ~5/6.

Story feting a macro HF trader with 18 straight years of positive performance from @nishantkumar07 grabbing attention this morning. Lets say his true Sharpe is 1.0 (good, not extraordinary), what are the odds of such a streak based simply on random chance? (don't cheat...)

@blockbaron69 @liquiditygoblin @weaponizedFOMO ofc lol. but saying “banks and mm” is saying “cheese and cheese”

@Solving_Inv @liquiditygoblin @weaponizedFOMO There are MMs that aren’t banks, hope this helps.

junior options interview question: you're on an index options market making desk & buy a front month straddle OTC with a bank, crossing at 24 vol. both you & the bank use the same option pricing model. you both think you made a good trade. why?

@liquiditygoblin @weaponizedFOMO i mean banks are mm so idk what your trying to get across

@Solving_Inv @weaponizedFOMO my follow up is just painting the hypothetical but I can see how the wording doesn’t help. Could you help suggest better wording?

@liquiditygoblin @weaponizedFOMO at first order i don’t understand how you difference MM and banks. And my answer is in response to your follow up not ur initial question

@Solving_Inv @weaponizedFOMO the point has nothing to do with banks or MMs behaviour & it is just that two different realised hedging strategies can yield different results. If your takeaway is “banks behave like X & market makers behave like Y” I wouldn’t have passed you on the interview.

@liquiditygoblin @weaponizedFOMO there

@Solving_Inv @weaponizedFOMO these are two participants & in APAC markets, some banks even hedge only wrekly

@liquiditygoblin @weaponizedFOMO since when does bank hedge once daily?

@weaponizedFOMO bank hedges once daily on close, mm hedges continuously ~30minutely. MM realises at 26vol, bank realises at 22vol both are happy

@idro___ fireducks?

Started to use polars, this is good and addictive

@quant_arb -matplotlib +plotly -pandas + fireducks

Only “quant specific” library Ive ever used were: - pyvollib vectorised for fast greek calcs - quantstats which is awful bc it requires an old version of pandas to use BUT it has a nice function to make a full html tear sheet of a strategies performance stats so I’ve used that - CCXT for when I’m too lazy to properly code up websockets (also useful to get all the min trade sizes etc from their get markets function) - tardis_dev library for data scraping - maybe in the odd instance I’ve used an exchange SDK library because I was either too lazy or needed a correct reference And all of these are fairly rare cases compared to the normal libraries. Otherwise it’s fairly normal stuff: - scipy - numpy - pandas - matplotlib - cvxpy - statsmodels - sklearn There’s not really any “special quant libraries” there’s a couple odds useful ones for when you’re too lazy to do it yourself and the vast majority are total slop. Zipline, pyfolio, openBB are not regularly used by any real quant and even the ones I listed it’s uncommon that I pull them out. They’re not in my default imports so to speak

@JonathanRemy8 alright, risk on, usdcad back down

@Solving_Inv Long cup&handle breakout

@quant_arb try 8

@Solving_Inv i wish i had at least 2 tb

If your RAM usage doesn't look like this then you aren't doing real research

@longironcondor @Biggiethelad1 Fireducks

@Biggiethelad1 Reasonable

@macrocephalopod Would u say the same for the comp. fin. masters at oxford?

If you want to land a job at a tier 1 firm, the best advice for pre-college students is to go to the most selective school that will accept you, and major in a hard subject like the ones listed above. If you’re already at university but not somewhere that’s a target, you need to do one of three things — 1. Be prepared to work *extremely* hard on your applications, grind interview prep, play the numbers game. 2. Moderate your expectations down, maybe your first job won’t be your ideal one but you can try to trade up if you work hard and excel. 3. Apply for masters or phd at a top school in a *selective* program in a hard subject, like the ones listed above. This rules out nearly all “quant finance” or “mathematical finance” or “financial engineering” masters programs.

The bigger problem is that the “masters in quant finance” programs are mainly revenue-generating machines for universities that accept students based on ability to pay rather than on merit. They are mostly not realistic routes to a job in trading or quant at a tier 1 (or even tier 2) firm, who still overwhelmingly prefer to recruit from undergrad, masters and phd programs in math, computer science, statistics or hard sciences.

@DannyDayan5 @jhvaranda @JackFarley96 @bennpeifert Momentum wins!

@DannyDayan5 @jhvaranda @JackFarley96 @bennpeifert what a deep conversation on mean reversion vs momentum

Emergency volatility reminder from @bennpeifert: Selling vol at 40 is not necessarily better than selling vol at 10! "Markets aren't that dumb." Full interview: Apple shorturl.at/ATujV Spotify shorturl.at/cvWHx