Delegates Please Publish Feeds More Often

[xeroc]

I finished it today:

Please feel free to check if correct!

https://github.com/xeroc/pytshares/blob/master/btsx_feed.py

[emski]

How can I check which delegate is doing his job properly and who isn't?
How do I check what feed every delegate publishes?

Basically in simple terms and assuming GUI and windows how can I check the delegates?

I hope BitShares Blocks can gather feed publishing statistics for all delegates including which feeds and frequency of update.

I've added the frequency of feed updates to the delegates info, it's calculated as the average of the number of feed updates over the last 48 hours. Multiple feed update transactions in the same block are only counted once, but I do count updates that are separated by a block or more for now. I can probably get update frequency for each asset added to the delegate page as well, will check that tomorrow.

I picked "updates/day" as the unit as it gives the "nicest" numbers, how many updates per day do you think is good enough?

The best metric I can think of is the average of the delegate feed error. You could pool xeroc's delegate feed difference each hour and provide the average.
However it might not be the "best" as some delegates have different policies. If that is within 2,3% it should be OK. Frequency of the updates is also good.

[bytemaster]

You don't want to publish the average "error" because then that will motivate everyone using the same feed and updating at the same time.    We want diversity of opinion, we just want it more frequently to adjust to changing market conditions.

[emski]

You don't want to publish the average "error" because then that will motivate everyone using the same feed and updating at the same time.    We want diversity of opinion, we just want it more frequently to adjust to changing market conditions.

Well frequency of updates is one thing. However if most of delegates have publicly available policies you can implement them and check the error according to their policy. And you can have your own policy and check delegate's error according to that. I'm not sure frequency of updates is the best metric as one might frequently publish "wrong"/questionable feeds.

[iHashFury]

Hello

I have just set-up my price feeds https://bitsharestalk.org/index.php?topic=7787.msg159042#msg159042

I hope its correct.

Feedback is appreciated.

[alt]

I saw there are 11.3 M USD margin call, but the feed price almost 10% higher than other exhchange. and change very slow.
if the witness still behave like this, the USD will be blackswan more easy.

if  REAL_PRICE < MEDIAN and YOUR_PRICE > MEDIAN publish price

When attempting to write a market maker the slow movement of the feed can be difficult.

I would recommend the following:

if  REAL_PRICE < MEDIAN and YOUR_PRICE > MEDIAN publish price
if  REAL_PRICE > MEDIAN  and  abs( YOUR_PRICE - REAL_PRICE ) / REAL_PRICE  > 0.005 publish price

The goal is to force the price down rapidly and allow it to creep up slowly.

By publishing prices more often it helps market makers maintain the peg and minimizes opportunity for shorts to sell USD below the peg that the market makers then have to absorb. 

If we can get updates flowing smoothly then we can gradually reduce the spread in the market maker bots. 

*note: all prices in USD per BTSX

[sahkan]

I have read through this and I agree. I have adjusted accordingly and sped up the price publishing to every 20 minutes. (This is of course the complete opposite of what bitcrab was asking for BitCNY).

[R]

Would it also not be worth increasing the max force settle volume per 24hrs? If it's set to 1% and someone has a margin call greater than 1%, perhaps the DEX couldn't clear the margin call fast enough?

[biophil]

Would it also not be worth increasing the max force settle volume per 24hrs? If it's set to 1% and someone has a margin call greater than 1%, perhaps the DEX couldn't clear the margin call fast enough?

I agree that max force settle volume is WAY too low. But I sort of assumed it didn't affect margin calls.

FWIW, the DEX doesn't clear margin calls -- the traders do. A possibly-bigger problem is the max short-squeeze ratio is only 110%, which means that margin calls are not very aggressive and traders don't have much incentive to buy into the margin call. I suppose USD will probably black-swan today because of it. A very serious problem is that Bitshares is too shorter-friendly, at the expense of people who hold the supposedly risk-free bitassets.

[biophil]

I have read through this and I agree. I have adjusted accordingly and sped up the price publishing to every 20 minutes. (This is of course the complete opposite of what bitcrab was asking for BitCNY).

Is bitcrab the one who thinks that bitasset risk should be moved from the shorts to the longs?

[R]

Would it also not be worth increasing the max force settle volume per 24hrs? If it's set to 1% and someone has a margin call greater than 1%, perhaps the DEX couldn't clear the margin call fast enough?

I agree that max force settle volume is WAY too low. But I sort of assumed it didn't affect margin calls.

FWIW, the DEX doesn't clear margin calls -- the traders do. A possibly-bigger problem is the max short-squeeze ratio is only 110%, which means that margin calls are not very aggressive and traders don't have much incentive to buy into the margin call. I suppose USD will probably black-swan today because of it. A very serious problem is that Bitshares is too shorter-friendly, at the expense of people who hold the supposedly risk-free bitassets.

What would the effect of changing the max short-squeeze ratio be specifically? Would you desire it increase of decreased? All committee controlled assets use 110%

[binggo]

The root problem is not the feed price and the max short-squeeze ratio.

The root problem is the DEX is very very friendly to the shorter, when someone was forced liquidation, the rest of BTS should be withdraw at least 60%  as a punitive margin by the DEX.

[biophil]

Would it also not be worth increasing the max force settle volume per 24hrs? If it's set to 1% and someone has a margin call greater than 1%, perhaps the DEX couldn't clear the margin call fast enough?

I agree that max force settle volume is WAY too low. But I sort of assumed it didn't affect margin calls.

FWIW, the DEX doesn't clear margin calls -- the traders do. A possibly-bigger problem is the max short-squeeze ratio is only 110%, which means that margin calls are not very aggressive and traders don't have much incentive to buy into the margin call. I suppose USD will probably black-swan today because of it. A very serious problem is that Bitshares is too shorter-friendly, at the expense of people who hold the supposedly risk-free bitassets.

What would the effect of changing the max short-squeeze ratio be specifically? Would you desire it increase of decreased? All committee controlled assets use 110%

In my view (bear in mind that I haven't worked through all of the details here), it should be increased. As I understand it, 110% means that a margin call is settled at the price feed plus 10%, which means essentially that the shorter is assessed a fixed penalty of 10% for letting their collateral ratio get too low. Alternatively, BitAsset holders are offered a 10% premium to trade in their BitAssets for BTS. In wild market swings (especially when the price feeds are lagging the external prices), 10% may be too low to clear a high volume of margin calls and prevent a black swan.

I have a vague feeling that the reason the MSSR exists in the first place is to provide a measure of protection against manipulation. Unfortunately the effect of this is to transfer risk from BitAsset short positions to BitAsset long positions.

[biophil]

The root problem is not the feed price and the max short-squeeze ratio.

The root problem is the DEX is very very friendly to the shorter, when someone was forced liquidation, the rest of BTS should be withdraw at least 60%  as a punitive margin by the DEX.

Yes, that's exactly what I'm saying. If I understand it correctly, that's what the max short-squeeze ratio does -- controls the penalty associated with getting margin-called.

[yvv]

There is no need for short squeeze protection imo. When a margin call happens, the system should take all orders up to a limit defined by collateral.