Latest article out where we fix LTR and make it actually outperform the state of the art benchmark.
The current literature version massively underperforms benchmark so we modify the objective and scaling to develop a novel model which beats existing models.

Reward
like
Comment
Repost
Share

StatArb

06-23 23:11

Bank I applied to make an account with >2 years ago emailed me back today 😵‍💫

Reward
like
Comment
Repost
Share

StatArb

06-19 12:14

The average startup is “multi task harness rails for agentic workflows” and then it’s just a browser with ChatGPT built in

Reward
like
Comment
Repost
Share

StatArb

06-17 16:55

Part 2 out now. We find multiple >5 Sharpe signals with high correlations to forward returns, and establish a feature set for factor modelling in the next article.

1 Likes

Reward
1
Comment
Repost
Share

StatArb

06-16 16:24

People who can’t find the momentum factor in crypto posses something in practitioner terms referred to as a “skill issue”

MMT-5.63%

Reward
like
Comment
Repost
Share

StatArb

06-09 17:46

Doing HFT sims 😵‍💫😵‍💫😵‍💫

Reward
like
Comment
Repost
Share

StatArb

06-09 09:27

Forecasting:
5min and below - XGBoost
15min and above - Ridge
You can enhance ridge by fitting non parametric regressions per feature as an in-between. To reduce fitting you can add a hypothesis, and only fit if it validates.
This is typically the optimal model by horizon.

1 Likes

Reward
1
Comment
Repost
Share

StatArb

06-08 17:17

I remember I went to a conference once and a quant fund exec (non technical) told me the full list of all the alt datasets they used.
He obviously had no idea that probably should’ve stayed confidential as testing datasets is a lot of work and many of them were niche, but it saved me testing 20 something datasets!

Reward
like
Comment
Repost
Share

StatArb

06-03 18:56

Here's an example of how an alpha decays as we trade less and less liquid assets. This the pre-fees performance of an alpha cross_ex_std_1w.
It is the standard deviation of volume across exchanges for a given timestamp, and then averaged over the past week. Basically how well do exchanges agree, you can also use OI it is about 0.86 correlation if you swap it out for OI.
It is very clear that as we trade less liquid and higher spread assets where it becomes increasingly harder and harder to monetise the signal gets worse.
It starts out at 2 sharpe and ends off at 0.3 Sharpe for top 30 (by mar

1 Likes

Reward
1
Comment
1
Share

StatArb

06-02 18:46

Never price improve

Reward
like
Comment
Repost
Share

StatArb

06-02 16:17

In the latest article we break down exactly how to research and evaluate HFT alphas, and explain advanced techniques used by top trading firms.

Reward
like
Comment
Repost
Share

StatArb

05-29 09:43

I find HPCA is a v useful tool for feature selection. Makes the clusters where you need to choose between features much more clear than a correlation grid

Reward
like
Comment
Repost
Share

StatArb

05-25 02:09

Forgot goal #2

Reward
like
Comment
Repost
Share

StatArb

05-23 08:45

Unit tests are an admission that your code might have errors. Real manual HFT developers don’t use them

Reward
like
Comment
Repost
Share

StatArb

05-23 06:00

ChatGPT has allowed otherwise incapable people to generate pretty graphs and this has revolutionised the quant slop space

Reward
like
Comment
Repost
Share

StatArb

05-21 12:17

I have seen some commentary about orderbook vs RFQ. Some thoughts of mine as someone who’s quoted both:
1. RFQ is inherently taker only and as such you hit a limit with costs and cannot ever reach ultra low cost execution required for higher turnover strategies via making into positions.
2. Orderbook will generally be better cost for anything that isn’t explicitly ultra large size
3. RFQ can outperform in situations where the net risk of the position is significantly different than the gross risk. This is multi leg positions and options structures (you pay for the gross of the Greeks when exec

Reward
like
Comment
Repost
Share

StatArb

05-20 09:33

For those who are not good at coding looking to try out quant strategies, I think: is a great product for researching and deploying strategies. Early stage but very cool! Worth trying :)
[This is a genuine endorsement, I have no financial interest here]