Prices from dexes

Prices calculated from trading data
We created a table that creates price feeds based on decentralized exchange trading data. This table covers much more assets than prices.usd, since it covers all assets that are traded on any of the decentralized exchanges that are indexed in dex.trades.
Please keep in mind that this script can generate wrong prices in rare cases.
This table is very resource intensive and can therefore only be updated every few hours, please keep that in mind when utilizing it. **** Also the resolution is only hourly, so if you need minutely prices do refer to prices.usd.
This table currently only exists for Ethereum on our old database architecture.
How this works
The logic of how this table works can be accessed in our public github repo.
This script generates median hourly prices based on data from decentralized exchanges found in dex.trades. It will assign asset prices based on a trading pair which has a pricefeed in prices.usd.
Let's take the $SPELL/ETH Pool for example.
$ETH price is contained in prices.usd
$SPELL price is not contained in prices.usd
In order to get the $SPELL price, the script will dynamically calculate the price of $SPELL based on the price of $ETH that was exchanged for it.
e.g. 5 $ETH were exchanged for 1,086,083 $SPELL.
Dex.trades will assign a usd_amount to this trade based on the $ETH price data in prices.usd.
That usd_amount is $23,498.
5 * price of ETH (4.699,6) = $23,498
Calculating the price of $SPELL is now as simple as dividing the amount of tokens exchanged with the usd_amount recorded in dex.trades.
$23,498/1,086,083 ≈ $0,02163
We now have successfully calculated the price of 1 $SPELL.
In order to correct for extreme outliers and in order for this table to be performant the script then aggregates all recorded data into one median_price per hour.
Known issues
In rare cases this script will generate price feeds that are based on illiquid pairs and therefore report wrong data. This happens when all liquid trading pools of this token do not have a price feed in prices.usd.
An example of this would be $PLAY, a metaverse index from piedao. The liquid trading pair for this asset is $PLAY/$DOUGH. The "correct" price of $PLAY is represented in this pool, but the combination of dex.trades and prices.prices_from_dex_data are not able to pick up this price.
Instead, dex.trades will only have a usd_amount for illiquid pairs of this asset. In this case, the $PLAY/$ETH pool has trades once in a while and these will have a usd_amount in dex.trades. The liquidity of the $PLAY/$ETH pool is very low and it pretty much only consists of arbitrage trades. Therefore, the resulting pricefeed in prices.prices_from_dex_data is faulty since it depends on the usd_amount in dex.trades.
In order to check for this, you should manually verify the results of prices.prices_from_dex_data in order to make sure arbitrage trades do not disturb the price feed constructed. A simple way of validating that the script is working with the right pools is checking the sample_size column. If the number seems suspiciously low, the script probably doesn't pick up the right price.
In cases like this, you have to manually construct a price feed.
Outro
We are always looking to improve this table, if you have any ideas or comments don't hesistate to open a PR or contact us in our Discord.

lending tables

labels

Last modified 4d ago

Copy link

Edit on GitHub

Outline

Prices calculated from trading data

How this works

Known issues

Outro