Dhananjay Nene

Wannabe Algorithmic Trader. Ex Programmer. This stream for markets/trading related tweets only. (Main General Purpose Twitter Id: @dnene)

[Dhananjay Nene @d7yn]

₹200 cr loss due to a single fat fingered trade 😱😱😱🤯

https://t.co/H9DazUvXKL

[Dhananjay Nene @d7yn]

@ChanderBhatia01 Isn't "all decisions" too wide a scope. I often "trade" with a 5 min view with probabilities as the underlying articulation of a trade rather than fair value. There's a case for both based on trade horizon .. 5 mins or 5 years.

[Dhananjay Nene @d7yn]

@ChanderBhatia01 Isn't SL relevant primarily when using technicals and trading? In fundamental investing if the price drops far below fair value it's an invitation to buy, not sell, no?

[Dhananjay Nene @dnene]

My talk on algorithmic trading code snippets at @FnConf is online https://t.co/E0MDoV0JOy

[Dhananjay Nene @d7yn]

@mickey_trades I just find it far far better.

[Dhananjay Nene @d7yn]

Will be talking about my algo trading software next month.

[Dhananjay Nene @d7yn]

@mickey_trades Kotlin

[Dhananjay Nene @d7yn]

@SubhadipNandy16 @PaytmMoney Since you've already gone down the path of a box chart it could be a curious exercise to use the IQR to just completely remove outliers that are more than 1.5 IQR away from the IQR borders. https://t.co/D9HVZfsUTr

[Dhananjay Nene @d7yn]

1st, that's like clamping down on forestation because that leads to weeds. Shooting yourself in the foot/Cutting the branch you're sitting on are all idioms that come to mind.

2nd, tech has a natural force of its own. Like flowing water. It will find a way around. Chrome plugins

[Dhananjay Nene @d7yn]

SEBI seems to have a grouse against those (mis)selling algorithmic Trading packages, and in a typical quasi-governmental knee jerk reaction they seem to want to clamp down on algorithmic trading itself. Which is ridiculous for two reasons ...

[Dhananjay Nene @d7yn]

Also manual trading requires folks to focus on one or two symbols at a time. Because human brain cannot process that data fast enough. A high speed algo OTOH can analyse the entire market every second, thus improving scalability n/n

[Dhananjay Nene @d7yn]

I tweeted this sometime back. An explanation of the challenge. I saw folks using heuristics around implied volatility (IV) of an option. Why heuristics? Because it is very time consuming to compute. Because the computation does not have a closed form solution. 1/n

[Dhananjay Nene @dnene]

Into a computational challenge even more challenging than I was into couple of decades ago.

Evaluating a function without closed form solution. Kotlin -> java, Boxed numbers to primitives, inlining, declaring finals etc etc. Anything above 1 millisecond is too slow 😒

[Dhananjay Nene @d7yn]

Almost all retail algos focus on rather simple algos. Also they rely on trend analysis, price action, and/or market behaviour analysis (open interest etc). The bet I'm willing to make is add extremely high speed quant to that and you get a much stronger set of rules. 4/n

[Dhananjay Nene @d7yn]

@brijwhiz Two of us 😀

[Dhananjay Nene @d7yn]

This trading journey is turning out to be big unexpected bet the farm.

Know enough to be a profitable manual trader. But hate manual trading since it requires continuous monitoring and responding to external events. That just tires me. Not something that I can do regularly

1/n

[Dhananjay Nene @d7yn]

@IAmChetanAhuja It's monitoring a large number of variables with lots of rules based on

1. Math telling you there's an opportunity
2. Market behaviour telling you there's an opportunity
3. Trend analysis telling you there's an opportunity.

You work with all to maximise win probability

[Dhananjay Nene @d7yn]

So yeah .. I expect another 2-3 months before the algos start rolling out. After that either this was a huge failed experiment or it is kachhing time. n/n

[Dhananjay Nene @d7yn]

And that experience and insight is way too hard to code. So am stuck between manual I don't wish to do and automated which is itself exceedingly hard. Just the confidence that if someone can indeed automate it, I certainly can keeps me going. 3/n