Feature: Curve V2 Sim

[chanhosuh]

• Handle multiple coins (at least 3 because of tricrypto) #91
• Create SimCurveCryptoPool class (Stub implementation of the SimCurveCryptoPool #180)
  • price #178
  • get_input_amount #179
  • misc methods: prepare_for_trades, assets
• Grid class should be refactored with mixins to extend current functionality to cryptopools (Update param samplers #158)
  • integration exercise (Integrate SimCurveCyptoPool with ParametrizedPoolInterator #165)
  • bugs: vol_mult param in autosim no longer works as expected #168, Plotting fails when no variable params #169
• metrics framework needs updating: getters, PoolBalance needs to incorporate price, PoolValue needs to use xcp(D)
• Update relevant network modules (we have the subgraph module updated and metadata is used to instantiate the cryptopool)
• Update volume multiplier logic (Update volume limits #182)
• Pipeline alterations to accommodate initial pool setup (Cryposim hacks #200)
• Add v2 params into autosim argument parser
• Tune/adjust crypto sims to exhibit useful/realistic behavior

[allt0ld]

I tried to simulate two V2 pools, not knowing they were not yet supported. Thought I'd contribute my findings. TLDR: my specific errors result from the large difference between simulated Stableswap prices and actual (volatile) prices where the peg is not 1:1.

Both pools were basic two-token ones, crv-eth (symbol "crvCRVETH") and cvx-eth (symbol "crvCVXETH") on mainnet, using the following code:

import curvesim
res = curvesim.autosim("[symbol]")

No parameters were overridden or added. The output, which I've truncated, looked like the below (this is the cvx-eth pool)

Warning: Opt_arb error, Pair: (1, 0) Pool price: 0.9996 Target Price: 0.003889433647690374 Diff: 0.9957105663523097
Warning: Opt_arb error, Pair: (1, 0) Pool price: 0.9996 Target Price: 0.003916063090355466 Diff: 0.9956839369096446
Warning: Opt_arb error, Pair: (1, 0) Pool price: 0.9996 Target Price: 0.003888620809118686 Diff: 0.9957113791908814

Naturally, the simulated pool is behaving as if both tokens were stablecoins under Stableswap (pool price close to 1), since only V1 pools are supported at the moment. The specific error is a ValueError handled by the except clause on line 658 in curvesim/pipelines/arbitrage.py, in function get_trade_args. Possible reasons:

• On lines 653-655 opt_arb(get_bounds, error_function, in_index, out_index, price) propagates a ValueError from subsequent calls in the call stack
• On line 656, min(trade[2], limit) raises a ValueError due to taking in an empty iterable with no default placeholder specified. Source

The error_function passed in is post_trade_price_error(dx, i, j, price_target) defined on line 74 of curvesim/pool/sim_interface/pool.py, which calculates dydx - price_target. It doesn't look like root_scalar(error_function,...) imported from scipy.optimize (line 476 of arbitrage.py) or error_function itself return a ValueError. Therefore, it may be possible to reuse the arbitrage pipeline if you redefine the invariant to suit V2, and modify the logic where you raise an exception based on Pool Price - Target Price, wherever it is.

[nagakingg]

Thanks @PhilipLu97, that sounds about right. The arbitrage pipeline should work once we make sim_interfaces for v2, which will provide the error_functions, etc.

Once we merge the new metrics/plotter, we will also need add entries to compute the metrics for v2 pools.

@chanhosuh -- So we don't forget, we also need to make bonding_curve and order_book support v2

[chanhosuh]

Epic finished with the release of 0.5.0.b1, although we will of course wrap up the data provider issues etc.