Ezon (@fryderyk) • Hey
Crypto HFT Quant Researcher | MEV searcher
Sharing my learnings.
Publications
- falshbots is a scam
- is there any defi community here?
- long time no lens
- Happy Easter 🌸
- wow my @orbapp.lens experience is so good, congrats team
- What is your favourite @lensprotocol app?
- deep appreciation and admiration for single moms 🙏
- wondering why on-chain trade execution needs to be so complicated - life would be so much easier if those things were implemented on the layer level
rn if you want to build an on-chain trading operation you literally need to have some good network/secret knowledge to avoid dumb mistakes and losing all your money
compare to cexes, how stupid
- Why is @lensprotocol better than Farcaster?
- What could we do to make @lensprotocol more popular?
- https://www.bbc.com/news/technology-64954119
Now we can't put more than 5k a month on chain? That's weird, no?
- Wondering what are the next big milestones for @lensprotocol
- What's the best city for degens? Why?
- gm
- are there any fitness influencers on lens?
- What's your favourite crypto conference?
- optimism will be new cosmos
- MEV is becoming the new HFT
Thoughts?
- gm what's your best arb so far?
- What's your lens score? 👀
- What is your favourite advantage of lens over twitter?
- What's one advice that changed your life?
- Let's take lens to the next level
- Best NFT collections to follow?
- What's your favourite NFT?
- Hardware wallet structure I recently considered:
1. "hodl" address - for holding only, no trading
2. "final trade" address. One directional transactions going straight to the hodl address.
3. "trade" address. For trading the assets on most trusted sources
4. rest, anything
If we split (1) into more addresses, we can exponentially increase our security against **some** attacks. Remember, security is key.
Thoughts?
- Sometimes you're doing the right thing but you're just not doing it for long enough.
Happened to me **a lot**.
Always make sure your sample size is large enough.
Thoughts?
- Quant research as web dev:
backend - data engineering
frontend - data science/ml/trading
devops - devops
one **subtle** difference is that a good frontend is earning 10x more
- Today in Poland we are celebrating "**Fat Thursday**"
All the shops are full of donuts and in the morning you see people saying "20 donuts please".
Much much better than some Valentine's day lol
- Reading "Machine Learning for Algorithmic Trading" by Jansen rn.
Will share what I think after I finish, looks good and beginner-friendly so far
- Do you think you have some good insight about crypto? Doesn't need to be technical!
Happy to help you convert it into a trading strategy 🙂
Just dm and follow!
- How to build Momentum/Pairs Trading?
There are 4 key compontents
- Opportunity discovery engine
- Good execution engine (preferably fast)
- A nice trading UI
- A good risk management (health monitoring, kill switch)
Follow for more trading things
- simple yet powerful
(bollinger bands)
- Is there anyone (apart from me) on @lenster.lens posting about algo trading/mev?
- "MEV is hard math"
bs, 90% of the time is devops
- Best way to get into trading/mev is the same as with anything else - just build something. It doesn't need to be perfect. *It will never be perfect*. Don't start with complex textbooks and don't search for a perfect strategy. Allow yourself to make mistakes and just **get started**.
Same with the gym. Many people go, overspend for an ideal outfit and lots of supplements. These things matter only later, at a higher level. **The first thing you should do is just go there, do something and be consistent**. That's it.
-
**Basic trading strategies, part 3**
*Pairs trading*
Let's start with some intuition: the classic example is Coca-Cola and Pepsi. You can guess that these stocks behave kinda similarly, so we could assume that their ratio is more or less constant. I.e. if the ratio grows - we can expect it to go back down. What do we do?
Yes, we short the overpriced stock and we long the underpriced.
This can be nicely generalised. I.e. instead of the ratio - you can take any function (i.e. you can take the difference and then trade the spread, but you can also be more creative). Instead of a pair you can take more securities, whatever.
Cool, so we have some intuition. Now let's find some pairs. The classic example is to test for cointegration (https://en.wikipedia.org/wiki/Cointegration). How do we do that? Well, there are lots of tests: i.e. Engle-Granger, Johansen or Phillips-Ouliaris.
Then let's just pick some nice tokens and test them in pairs (dm if you want to see the code). We'll calculate the p-values. The lower - the more likely that we have a cointegrated pair. Below is the heatmap with the p-values
You could say that aave and atom have a pretty low p-value, hm? Let's take a closer look
Let's take a look at the second graph below - the graph of their ratio (actually zscore). We could have a look at our previous post and think of some mean reversion here. BUT, in fact, things get very bad when we increase the timeframe.
The third graph is how bad it gets.
In general, pairs are great, especially in crypto. One example thing that could be useful to do before is applying Kalman filters. We'll talk about this in some future posts.
Follow for more crypto/algo trading/mev strats!
- @lenster.lens feels so clean compared to twitter
- **Basic trading strategies, part 2**
*Mean reversion*
Mean reversion is a popular trading strategy that is based on the idea that prices tend to return to their historical average over time (note how this is opposite to the momentum from the last post).
Let's start with an example. On the graph below is a price of some weird coin (guess which one) over a few years and 2 moving averages: 25-day and 50-day.
The strategy could be as follows:
-when the short-term moving avergage intersects the long-term average and goes below it we go long (because we assume it will revert)
-otherwise we go short
The graph perfectly shows how **tragic** this strategy is during bull runs (notice how we short at the first intersection). However, in bear market (or when there is less momentum), this strategy can be very profitable. Ofc, it's very important to adjust the timeframe and manage risk properly. Properly balancing mean reversion and momentum can be alpha
Overall it's quite hard to find good mean-reversion based strats, *very alpha*
Follow for more crypto/algo trading/mev strats
- **Basic trading strategies, part 1**
*Momentum*
The underlying principle of momentum trading is that securities that have been rising in price are likely to continue to rise, while those that have been falling in price are likely to continue to fall.
Then, for instance, we can open a long position in a security that is showing positive momentum or a short position in a security that is showing negative momentum. Ok, but how do we measure momentum? There are many ways.
One example is Moving Average Convergence Divergence (MACD). In MACD we basically compute two different moving averages and look for when they intersect. Those points then indicate the direction of our momentum (pic below).
On the graph below, at time ~600, the intersection indicates negative momentum. The next one is positive and so on.
Now, there are many ways in which we can use momentum. An interesting one is ranking securities and trading outliers (more about this in a separate post).
Overall, momentum trading is a great strategy, but it also involves a high level of risk (ofc there are many things out of our control, like macro events).
Follow to learn more about crypto/algo trading/mev
- siemano