The importance of Meta Aggregators on top of DEX aggregators

What's the point of aggregating aggregators?

That's the question everyone asked two years ago when the first iterations of Meta Aggregators appeared.

Many saw it as a superfluous add-on in crypto, given that liquidity aggregators, such as 1inch or 0x, would offer a similar and interchangeable service.

Today, we are digging into the data to see that this statement does not hold and that Meta DEX aggregators like Jumper.exchange or DeFillamaswap are the most suitable option to cover the full scope of trades by types of pairs and trades.

Let’s recap those three concepts for newbies:

A decentralized exchange (or

DEX

) is a peer-to-peer marketplace where transactions occur directly between crypto traders. This marketplace is orchestrated by a set of smart contracts where some users (liquidity providers) pool their liquidity and enable other users (traders) to swap tokens.

Example: Curve Finance, Uniswap, Balancer.

A DEX aggregator

takes the concept of DEX a step further.

This allows users to access a wider range of trading pairs and liquidity pools without needing to navigate multiple DEX separately.

DEX aggregators often offer improved price execution and reduced slippage by routing trades, and optionally splitting them, to one or a combination of several liquidity sources

Example: 1inch, 0x, Odos, Jupiter.

A Meta aggregator

builds upon the functionality of several DEX aggregators and acts as a route comparator to give the best option to the user.

Contrary to DEX aggregators, it does not split trades in a combination of multiple routes, it just chose the best one.

How are we digging into the data?

What we want to assess in this article is the trade efficiency given specific parameters.

Trade efficiency is defined as the ratio between amountOut (i.e. the amount received by the trader) and amountIn in USD.

The scope of the analysis is as follows:

This analysis will not take into account the potential time variation of DEX aggregations as data points are only collected for the last month.

Also the trades prices in USD are defined by the prices.usd table on Dune analytics: (https://docs.dune.com/data-catalog/spellbook/top-tables/prices#pricesusd)

Given that prices are aggregated among several sources of truth, we assume that its impact is negligible over hundreds of thousands of transactions.

We end up with ~504,000 observations:

~250k from 1inch trades, ~150k from 0x and ~100k from Odos.

In order to understand if the trade efficiency in price is dependent on the variables I described in the above section, I am using an Kruskal-Wallis test:

“The Kruskal–Wallis test is a statistical test used to compare two or more groups for a continuous or discrete variable. It is a non-parametric test, meaning that it assumes no particular distribution of your data and is analogous to the ANOVA. If the p-value is <0.05 the variable tested is likely to be significant”

ELI5: if p-value is smaller than 0.05, then DEX aggregators are not similar you dumm dumm

When applying this test over DEX, the p-value is 0. It strongly indicates the significant difference among DEX for trade efficiency:

This finding alone proves that DEX aggregators are significantly distinct from each other in terms of their price efficiency.

With also apply the same test but with the type of pairs:

And to the bucket size:

Given the really low p-value, those variables are significant in explaining trade efficiency.

After this first analysis, let visualize specificity of DEX aggregators.

0x protocol trades tends to be more performant in low-size trades (up to 10k) than 1inch, and then we see the opposite for high-value trades (10k-1m+).

Odos seems to be super consistent along any size of trades, but is mainly due to the fact that most traded pairs are stables.

When we remove the stable pairs from the equation for 0x protocol, Odos and 1inch, Odos proves to be a bit more volatile and consistent than others on each and every pair.

When we remove volatile pairs from the equation, each of the DEX aggregators seem to have a similar performance.

On volatile pairs, DEX aggregators keep getting different behaviours.

Odos and 0x trade efficiency deteriorates for bigger trade size (while Odos is consistenly better than 0x), while 1inch seems to improve on trade between $10k and $100k.

When we look by type of traded pairs, we see that in general 0x and Odos offer more regularity and better prices on average than 1inch on alternative pairs.

As seen before, stable pairs seems to be much more similar between each other.

This short demonstration aims to show that DEX aggregators have their own specificity. Some are more efficient for specific trade sizes or certain types of pairs.

From a user perspective, it may be more interesting to use these meta aggregators in order to have a 360-degree view of the quotes offered and be able to execute trades under the best possible conditions.