In my algorithmic trading company, Meru Capitals, the 4 of us manage captive funds, and work hard to generate consistent returns while simultaneously operating at scale. All our trades are automated i.e. generated and executed using algorithms generated in-house. A thread.. by Saurav KediaWe create trading algorithms (algos) which are computer programs that monitor market quotes and generate trading signals using embedded logic. These signals are then traded automatically by second layer called execution engine. All of this is done without any human intervention.
Our current capital base is ~ 40 crs and largely deployed in index options and mostly on short (selling) side. We look the generate 25%+ annual profit. In trading, Drawdowns (DD) loosely means losses made from from Peak profits level. We try to keep DD within 5%.
We are very focused on keeping the DD low. We want to minimize volatility and increase reliability in performance and in the process, ended up trading only intraday algos and that too the ones which place a large number of orders. Both of them improve reliability.
Such trading style also necessitated us to completely junk cash segment in 2021, as the trend following nature of those algos led to unacceptably high 15% DD in May. We moved fully into index options, as risk return profile is more stable there.
Within options, we started out with popular algos like 9.20 straddles, but we stopped them by Nov 2021, as we could see that eventually they will also generate high DDs. A 9.20 straddle variant (algo: IOTB3) below which we stopped in Nov'21. Look at its profitability before Nov.
We eventually moved on to creating algos which are continuously trading in the market and seeks to profit from a mix of theta decay and directional moves. We place very small orders and like to spread them through out the day, again with a view to minimize DDs. Once such algo:
In this options space as well, we look to diversify through mix of trend following/mean reverting algos as well NIFTY/BANKNIFTY algos. Currently we trade upto 15+ algorithms.
We pick up weights for individual algos in portfolio using a mathematical process of creating mean variance portfolios using Monte Carlo Simulations, again to reduce DDs.
This process has worked fairly well for us and YTD (since Jan), we are up roughly 20% with a max DD of 2.9%, in a market which was marked by excessive volatility triggered by rising inflation, Russia Ukraine war and Fed tightening. Thats > 5.5 crs of profits after charges.
Execution is the holy grail. All this requires strong software capabilities, and we have developed an in-house execution software: Kanhoji. Kanhoji is a modern software designed using distributed event driven architecture much like modern e-commerce apps and is highly scalable.
The name Kanhoji itself is inspired from mighty Maratha naval lord Kanhoji Angre, and we see ourselves as proud pirates holding up in a sea overrun by mighty institutional players with armies of Ivy league trained soldiers.
Since Jan 2022, we have placed 1.5 million orders with brokers and written premia worth Rs 800crs, thats 2.5 mln lots. On a single day, we have traded even 25K+ positions (full trade with entry and exit). All of these is managed by Kanhoji and is overseen by our 4 member team.
It also has an admin panel which allows us to handle administrative tasks like defining strategy parameters, capital allocation to various strategies, broker details etc.
Kanhoji also supports a fully featured back-office which generates performance reports and equity curves for us alongside various other reports including back-office reconciliations.
Kanhoji runs on AWS cloud infrastructure where we run a 4-machine configuration. Effectively, Kanhoji doesn't require us at all. All of us can be away for a day and it will trade for us. But we do keep coverage to avert any potential faults.
Consequently, most of our time is spent in developing new algos and improving Kanhoji further i.e. building for future.
The opportunities in this space are vast and we continuously look to collaborate with others: traders, researchers, brokers, and FIs. We want to trade new markets like currencies, commodities and looking to speak to others who already trade them.
A chance encounter with ex-colleagues led to discussions around what I really do. This is one question which I repeatedly get asked, especially from family. This is a specialized area which most don't know of. Many think I am a wild speculator and carry a lot of misconceptions.
Our style of algorithmic intraday trading is speculation all right, but well-thought, process-driven, rules based and very disciplined.
Thank You Saurav For this very informative thread.
Love,
DPZ.
Comments
Post a Comment