I recently saw a tweet where Nithin Kamath mentioned that building an OMS is quite difficult and that they still rely on third-party vendors. Honestly, that was shocking to hear from a brokerage that’s been around for two decades. The real issue, in my opinion, is not difficulty but mindset. Indian brokers simply aren’t willing to spend serious money on top-tier engineers. In the HFT world, small teams of highly skilled engineers build entire low-latency, highly reliable trading platforms from scratch. If that’s possible, why is it considered “too hard” for retail brokerages? It feels like brokers prefer to cut costs, rely on vendors, and market their platforms as “best in class,” while traders continue to suffer from latency, outages, and reliability issues. What do you think—tech limitation or cost-cutting disguised as pragmatism?
Here’s a topic-thread with the details from the rest of the tweets in the twitter-thread.
FWIW, Indian brokers ARE building their own OMSes at their own pace,
and will likely prioritize it higher / invest more in an in-house OMS / make it their primary system,
if/when there are any indicators of increased likelihood of such an in-house OMS offering better returns to the broker in the form of:
- higher customer acquisition
- higher customer-retention (better customer experience)
- lower operational costs (infrastructure, compliance, risk-management)
- lower risk of being disrupted by competitors (existing or new ones)
NOTE: Not just non-zero returns, but returns higher than the other areas where the brokers are expending (and will have an option to expend) their capital/time…
…all towards increased short-term or long-term profits for the broker.
2 dissimilar problem statements, with different constraints, will likely have 2 different optimal solutions.
-
A smaller limited problem statement, with limited liability, and easier to manage risk due to far better control over the users, (in-house HFT)
-
A different larger problem statement with external users (adversarial even?) with a wide variety of distributed users, with their distinct use-cases, and usage-patterns (general OMS for a retail broker).
So what is an individual retail trader going to do?
Bear the cost of switching their broker,
only to face similar issues (or possibly even worse) from another broker? 
Neither. Capitalism working as intended.
Note: Not claiming that it is in the best interests of the individual retail trader.
An alternate approach offering lesser demonstrable returns than the current status-quo?
= Status-quo continues.
Just the usual incentives -
Presumably there are other areas that a broker can focus on that offer better return-on-investment (RoI), key differentiation from competitors (Moat), require lesser upfront investment, with higher likelihood of returns than developing and maintaining their own OMS.
Here's a related thought experiment (directed at the individual retail trader)
Note: The following is an attempt to employ the same language as the previous post, but in the context of an individual retail trader, to hopefully make apparent the challenges with the proposed alternative.
Why do experienced Indian traders still depend on retail brokers after years of trading?
I recently saw a post on the TradingQnA forum that mentioned that running a brokerage is difficult and that individual retail traders still rely on third-party retail brokers who suffer from latency, outages, and reliability issues. Honestly, that was shocking to hear from traders managing significant capital over several years.
The real issue, in my opinion, is not difficulty but mindset. Indian traders simply aren’t willing to invest a fraction of their profits into professional infrastructure. In the HFT world, no one relies on a generic mobile app or a shared connection used by a million other people. If that’s possible, why is it considered “too hard” for the individual retail trader?
It feels like individual retail traders prefer to cut costs, rely on brokers, while continuing to risk lakhs on slippage, frozen terminals, and broker outages. What do you think - tech limitation or cost-cutting disguised as pragmatism?
Confronted with the above thought, individual retail traders will continue to focus on their strategies, while relying on their brokers for execution, who in turn will rely on other 3rd-party systems, built on top of other shared infrastructure, …
In the modern world, with all the complexities in most domains, complete end-to-end vertical integration is impractical for individuals and most corporations.
PS: This is a very common theme in the software industry. Additional discussions with nuances on this theme can be found by searching for and reading about “Build vs. Buy”.
Zerodha has been pretty stable for few years now. Rarely there have some delay issues but overall they are fine.
Having an inhouse OMS under development probably gives them some protection against bad terms ( ex i think MCX had some issues with vendor), but deploying and migrating it will be risky with so many users + existing solution has been pretty stable.
Hopefully when the transition does come, they can do it without major issues.
If you have any idea about OMS it’s not about tech. It’s about experience so it can be build with time only.
Dhan has their in-house DEXT OMS which consistently puts them at the top of latency benchmarks 
I have run a production intraday algo on it with serious capital, execution is fast 
but due to its parallel nature it takes about 0.5s for the RMS to be synchronized after an order execution. 
I never faced latency issue with zerodha from day one .
I don’t think any broker can claim their full Oms is in house if it was so issue all brokers would have built it .
Full times brokers still use TCS built websites not sure why .
They are fast, but Groww and Finvasia are faster as per openbroker.in in terms of response times.
Dhan OMS also brings in other constraints in terms of margin release times (or the sync) where they are the slowest - which is why basket orders are a hassle with them (they claim to have solved this very recently but it still has issues). I was advised to leave a time delay of atleast 1 sec (was more 1.5 years back) btw API orders for their margining system to catch up Dont know if that’s changed now.
Whats your trading volume? If your turnover is couple of lakhs, then you’ll never know how bad the latency is.
Haven’t tried Groww, surprised about Finvasia/Shoonya as their web interface was too slow and clunky when I tried it in 2024 
. It was the only broker I signed up with and closed my account since I couldn’t imagine a scenario where I would be forced to use it. Maybe they have more enterprise-y products with better response times? 
This has improved, I was told to use a 0.5s delay and it worked for the few months I ran a strangle algo back in Q1’25
Yes, clunky as hell… But their API is efficient enough 
I dont use them nowadays.
Trust (the perception of it) is everything 