App::cryp::arbit::Strategy::merge_order_book - Using merged order books for arbitration
This document describes version 0.010 of App::cryp::arbit::Strategy::merge_order_book (from Perl distribution App-cryp-arbit), released on 2021-05-26.
In your cryp.conf:
[program=cryp-arbit arbit] strategy=merge-order-book
or in your cryp-arbit.conf:
[arbit] strategy=merge-order-book
This is actually the default strategy, so you don't have to explicitly set strategy to this strategy.
strategy
[arbit-strategy/merge-order-book] ...
This arbitration strategy uses information from merged order books. Below is the description of the algorithm. Suppose we are arbitraging the pair BTC/USD. E1, E2, ... En are exchanges. P* are prices. S* are sizes. i denotes exchange index.
First step: get order books from all of the involved exchanges, for example:
# buy orders on E1 # sell orders on E1 price size price size ----- ---- ----- ---- P1b1 S1b1 P1s1 S1s1 P1b2 S1b2 P1s2 S1s2 P1b3 S1b3 P1s3 S1s3 ... ... # buy orders on E2 # sell orders on E2 price size price size ----- ---- ----- ---- P2b1 S2b1 P2s1 S2s1 P2b2 S2b2 P2s2 S2s2 P2b3 S2b3 P2s3 S2s3 ... ... ...
Note that buy orders are sorted from highest to lowest price (Pib1 > Pib2 > Pib3 > ...) while sell orders are sorted from lowest to highest price (Pis1 < Pis2 < Pis3 < ...). Also note that P1b* < P1s*, unless something weird is going on.
Second step: merge all the orders from exchanges into just two lists: buy and sell orders. Sort buy orders, as usual, from highest to lowest price. Sort sell orders, as usual, from lowest to highest. For example:
# buy orders # sell orders price size price size ----- ---- ----- ---- P1b1 S1b1 P2s1 S2s1 P2b1 S2b1 P3s1 S3s1 P2b2 S2b2 P3s2 S3s2 P1b2 S1b2 P1s1 S1s1 ...
Arbitrage can happen if we can buy cheap bitcoin and sell our expensive bitcoin. This means P1b1 must be above P2s1, because we want to buy bitcoins on E1 from trader that is willing to sell at P2s1 then sell it on E1 to the trader that is willing to buy the bitcoins at P2b1. Pocketing the difference (minus trading fees) as profit.
No actual bitcoins will be transferred from E2 to E1 as that would take a long time and incurs relatively high network fees. Instead, we maintain bitcoin and USD balances on each exchange to be able to buy/sell quickly. The balances serve as "working capital" or "inventory".
The minimum net profit margin is min_net_profit_margin. We create buy/sell order pairs starting from the topmost of the merged order book, until we can't get min_net_profit_margin anymore.
Then we monitor our order pairs and cancel them if they remain unfilled for a while.
Then we retrieve order books from the exchanges and start the process again.
Order books contain information about prices and volumes at each price level. This serves as a guide on what size our orders should be, so we do not have to explicitly set order size. This is especially useful if we are not familiar with the typical volume of the pair on an exchange.
By sorting the buy and sell orders, we get maximum price difference.
Order books are changing rapidly. By the time we get the order book from the exchange API, that information is already stale. In the course of milliseconds, the order book can change, sometimes significantly. So when we submit an order to buy X BTC at price P, it might not get fulfilled completely or at all because the market price has moved above P, for example.
max_order_size_as_book_item_size_pct
Number 0-100. Default is 100. This setting is used for more safety since order books are rapidly changing. For example, there is an item in the merged order book as follows:
type exchange price size item# ---- -------- ----- ---- ----- buy exchange1 800.1 12 B1 buy exchange1 798.1 24 B2 ... sell exchange2 780.1 5 S1 sell exchange2 782.9 8 S2 ...
If `max_order_size_as_book_item_size_pct` is set to 100, then this will create order pairs as follows:
size buy from buy price sell to sell price item# ---- -------- --------- ------- ---------- ----- 5 exchange2 780.1 exchange1 800.1 OP1 7 exchange2 782.9 exchange1 800.1 OP2 ...
The OP1 will use up (100%) of item #S1 from the order book, then OP2 will use up (100%) item #B1 from the order book.
However, if `max_order_size_as_book_item_size_pct` is set to 75, then this will create order pairs as follows:
size buy from buy price sell to sell price item# ---- -------- --------- ------- ---------- ----- 3.75 exchange2 780.1 exchange1 800.1 OP1 5.25 exchange2 782.9 exchange1 800.1 OP2
OP1 will use 75% item S1 from the order book, then the strategy will move on to the next sell order (S2). OP2 will also use only 75% of item B1 (3.75 + 5.25 = 9, which is 75% of 12) before moving on to the next buy order.
Please visit the project's homepage at https://metacpan.org/release/App-cryp-arbit.
Source repository is at https://github.com/perlancar/perl-App-cryp-arbit.
Please report any bugs or feature requests on the bugtracker website https://github.com/perlancar/perl-App-cryp-arbit/issues
When submitting a bug or request, please include a test-file or a patch to an existing test-file that illustrates the bug or desired feature.
App::cryp::arbit
Other App::cryp::arbit::Strategy::* modules.
App::cryp::arbit::Strategy::*
perlancar <perlancar@cpan.org>
This software is copyright (c) 2021, 2018 by perlancar@cpan.org.
This is free software; you can redistribute it and/or modify it under the same terms as the Perl 5 programming language system itself.
To install App::cryp::arbit, copy and paste the appropriate command in to your terminal.
cpanm
cpanm App::cryp::arbit
CPAN shell
perl -MCPAN -e shell install App::cryp::arbit
For more information on module installation, please visit the detailed CPAN module installation guide.