Author Topic: poll for the percent based transfer fee  (Read 10450 times)

0 Members and 1 Guest are viewing this topic.

Offline bitcrab

  • Committee member
  • Hero Member
  • *
  • Posts: 1900
    • View Profile
  • BitShares: bitcrab
  • GitHub: bitcrab
poll for the percent based transfer fee
« on: January 20, 2016, 07:59:28 am »
percent based transfer fee will be implemented, https://bitsharestalk.org/index.php/topic,21080.0.html the next step will be determining the parameters. let's see what a parameter combination will be welcome, you can also suggest other combination if you do not like the listed ones.
« Last Edit: January 20, 2016, 08:07:32 am by bitcrab »

Offline clayop

  • Hero Member
  • *****
  • Posts: 2033
    • View Profile
    • Bitshares Korea
  • BitShares: clayop
Re: poll for the percent based transfer fee
« Reply #1 on: January 20, 2016, 08:10:31 am »
I asked [member=21903]roadscape[/member] to get historical data of transfer activities to estimate potential impacts of each possible scenario.
We may need more solid base to justify a certain fee structure.

Addition:
I think percentage based fee can satisfy both sides (micropayment vs. network profit/referral system). So I would suggest like,
1 BTS (lower than the current) / 100 BTS (higher than the current) / 0.1% (These are just my rough hypothetical numbers)
« Last Edit: January 20, 2016, 08:14:42 am by clayop »
Bitshares Korea - http://www.bitshares.kr
Vote for me and see Korean Bitshares community grows
delegate-clayop

Offline bitcrab

  • Committee member
  • Hero Member
  • *
  • Posts: 1900
    • View Profile
  • BitShares: bitcrab
  • GitHub: bitcrab
Re: poll for the percent based transfer fee
« Reply #2 on: January 20, 2016, 08:30:22 am »
I think percentage based fee can satisfy both sides (micropayment vs. network profit/referral system). So I would suggest like,
1 BTS (lower than the current) / 100 BTS (higher than the current) / 0.1% (These are just my rough hypothetical numbers)

I do not understand why you set cap to 100 BTS, which means each time one transfer more than 100k BTS will pay 100BTS, about $0.3
in my view the cap should be close or a little above 0.0001 BTC, that's why I prefer 20 BTS.
it will be funny if the BTS transfer be more expensive than inter-Bank CNY transfer.


Offline clayop

  • Hero Member
  • *****
  • Posts: 2033
    • View Profile
    • Bitshares Korea
  • BitShares: clayop
Re: poll for the percent based transfer fee
« Reply #3 on: January 20, 2016, 08:35:11 am »
I think percentage based fee can satisfy both sides (micropayment vs. network profit/referral system). So I would suggest like,
1 BTS (lower than the current) / 100 BTS (higher than the current) / 0.1% (These are just my rough hypothetical numbers)

I do not understand why you set cap to 100 BTS, which means each time one transfer more than 100k BTS will pay 100BTS, about $0.3
in my view the cap should be close or a little above 0.0001 BTC, that's why I prefer 20 BTS.
it will be funny if the BTS transfer be more expensive than inter-Bank CNY transfer.

They are hypothetical examples. The main idea is it can be higher than the current level.
Bitshares Korea - http://www.bitshares.kr
Vote for me and see Korean Bitshares community grows
delegate-clayop

Offline clayop

  • Hero Member
  • *****
  • Posts: 2033
    • View Profile
    • Bitshares Korea
  • BitShares: clayop
Re: poll for the percent based transfer fee
« Reply #4 on: January 20, 2016, 05:54:59 pm »
https://drive.google.com/file/d/0B9wS1MU8AsmNbUpaXzZid0JSWjQ/view?usp=sharing  (98M)

Based on some descriptive analysis, I think 0.05% / 1 Min / 300 Max is reasonable numbers.


P.S Special thanks to [member=21903]roadscape[/member] for providing the data!
« Last Edit: January 20, 2016, 06:06:31 pm by clayop »
Bitshares Korea - http://www.bitshares.kr
Vote for me and see Korean Bitshares community grows
delegate-clayop

Offline abit

  • Committee member
  • Hero Member
  • *
  • Posts: 4549
    • View Profile
    • Abit's Hive Blog
  • BitShares: abit
  • GitHub: abitmore
Re: poll for the percent based transfer fee
« Reply #5 on: January 20, 2016, 06:03:19 pm »
https://drive.google.com/file/d/0B9wS1MU8AsmNbUpaXzZid0JSWjQ/view?usp=sharing  (98M)

Based on some descriptive analysis, I think 0.05% / 1 Min / 300 Max is reasonable numbers.
Thanks for the analysis. Would you like to provide a summary of the analysis here? It's a bit hard to download a 98M file or open it in Google Sheets.
BTS account: abit
BTS committee member: abit
BTS witness: in.abit

Offline yvv

  • Hero Member
  • *****
  • Posts: 1186
    • View Profile
Re: poll for the percent based transfer fee
« Reply #6 on: January 20, 2016, 06:05:26 pm »
What fee are we talking about here? Trading fee or transfer fee?

Offline clayop

  • Hero Member
  • *****
  • Posts: 2033
    • View Profile
    • Bitshares Korea
  • BitShares: clayop
Re: poll for the percent based transfer fee
« Reply #7 on: January 20, 2016, 06:07:00 pm »
https://drive.google.com/file/d/0B9wS1MU8AsmNbUpaXzZid0JSWjQ/view?usp=sharing  (98M)

Based on some descriptive analysis, I think 0.05% / 1 Min / 300 Max is reasonable numbers.
Thanks for the analysis. Would you like to provide a summary of the analysis here? It's a bit hard to download a 98M file or open it in Google Sheets.

Yup that's my plan. Sometime today :)
Bitshares Korea - http://www.bitshares.kr
Vote for me and see Korean Bitshares community grows
delegate-clayop

Offline clayop

  • Hero Member
  • *****
  • Posts: 2033
    • View Profile
    • Bitshares Korea
  • BitShares: clayop
Re: poll for the percent based transfer fee
« Reply #8 on: January 20, 2016, 06:42:40 pm »
There can be another point, which has a higher fee percentage (over 1%) and higher upper limit. Still digging in the data.
Bitshares Korea - http://www.bitshares.kr
Vote for me and see Korean Bitshares community grows
delegate-clayop

Offline clayop

  • Hero Member
  • *****
  • Posts: 2033
    • View Profile
    • Bitshares Korea
  • BitShares: clayop
Re: poll for the percent based transfer fee
« Reply #9 on: January 21, 2016, 06:52:41 am »
OK, here'are some summaries of analysis.

1. Minimum fee does not significantly influence total fee. Given 1% and 300 max (as suggested by jakub), min fee change from 1 to 6 only increases 1.8% of total fee. So I easily conclude that minimum fee is not that important, and lower minimum fee is more preferable to encourage micro-transactions. However, the minimum fee should be efficient to prevent spam. Therefore, I fixed the minimum fee to 1 BTS.

2. Given 1 BTS minimum, I tested scenarios proposed by bitcrab and jakub (0.1%/20 and 1%/300). Bitcrab's was 137k BTS fee in total while Jakub's is 1,830k BTS. I think the former is too small compared to transfer fees collected with 30 BTS flat fee. (137k vs. 13,212k, roughly 1%). While, the latter is 14%.

3. I tried to find a way to have the same level of fee as actually collected. But all scenarios are less then the 13mil BTS. So I conclude that it is impossible to have equal amount of fee under percentage-based fee system.

4. I checked how many txs are supposed to pay more (over 30 BTS) under 1% fee. It is found that 45% (7317/16243) of all transfers are affected implying more burdens in general. However, the rate decreases to 30% with 0.1% fee, and 25% with 0.05%. Meanwhile, 55% of transfers are benefited with 1% fee system, 69% with 0.1%, and 75% with 0.05%. The ratio of minimum-fee-paying transfers are 37%, 50%, 53% respectively (under 1%/0.1%/0.05% fee system).

These are my findings so far, and the followings are my opinions and suggestions.

1. I tried to have 5% of total fee actually collected. That is around 600k BTS.

2. I compared two fee rates, 0.1% and 0.05%. With 0.1%, 150 BTS maximum has 593k BTS fee. With 0.05%, 300 BTS maximum has 615k BTS fee.

3. I think lower percentage+high maximum is preferred to higher percentage+low maximum, because it is easy to justify "the bigger, the more". Given 0.05%/300, transfers over 600k BTS ($1800) pay the maximum fee of 300 BTS ($1).

So my suggestion is 0.05%/1/300. If you want to analyze by yourself, https://cryptofresh.com/bts_tx_history.csv will be the starting point (Thanks Roadscape).


Something I made wrong significantly. (didn't remove non-BTS). I will make it again soon.
« Last Edit: January 21, 2016, 04:25:01 pm by clayop »
Bitshares Korea - http://www.bitshares.kr
Vote for me and see Korean Bitshares community grows
delegate-clayop

Offline abit

  • Committee member
  • Hero Member
  • *
  • Posts: 4549
    • View Profile
    • Abit's Hive Blog
  • BitShares: abit
  • GitHub: abitmore
Re: poll for the percent based transfer fee
« Reply #10 on: January 21, 2016, 08:19:44 am »
Thanks for your hard work [member=19219]clayop[/member] .
One more thing need to be considered add into analysis: If we apply % fee to an asset, hopefully quantity of micro payments in that asset will increase, and perhaps quantity of big payments will decrease. If we make an estimation for example apply *5 to the former and 1/3 to the latter(depends on the parameters change), how will the result change?

//For example, 30BTS is unacceptable for most of tipping UIAs, so there are very few tips in the historical data. If we set the fee to 1BTS, perhaps tipping would get much more active.
« Last Edit: January 21, 2016, 08:22:28 am by abit »
BTS account: abit
BTS committee member: abit
BTS witness: in.abit

Offline clayop

  • Hero Member
  • *****
  • Posts: 2033
    • View Profile
    • Bitshares Korea
  • BitShares: clayop
Re: poll for the percent based transfer fee
« Reply #11 on: January 21, 2016, 08:29:34 am »
Thanks for your hard work [member=19219]clayop[/member] .
One more thing need to be considered add into analysis: If we apply % fee to an asset, hopefully quantity of micro payments in that asset will increase, and perhaps quantity of big payments will decrease. If we make an estimation for example apply *5 to the former and 1/3 to the latter(depends on the parameters change), how will the result change?

//For example, 30BTS is unacceptable for most of tipping UIAs, so there are very few tips in the historical data. If we set the fee to 1BTS, perhaps tipping would get much more active.

I will do it when I have time. But I think big transfers won't decrease much, because they are real demands. On the other hand, we can expect more micro-transactions, but I cannot guess the number because if some services using micro-payment (e.g. messaging / telecommunications) are really successful, they can increase the number dramatically. Assuming 10,000 messages are sent a day, the total fee goes up from 600k to 1.1mil BTS.
Bitshares Korea - http://www.bitshares.kr
Vote for me and see Korean Bitshares community grows
delegate-clayop

Offline abit

  • Committee member
  • Hero Member
  • *
  • Posts: 4549
    • View Profile
    • Abit's Hive Blog
  • BitShares: abit
  • GitHub: abitmore
Re: poll for the percent based transfer fee
« Reply #12 on: January 21, 2016, 08:37:25 am »
Thanks for your hard work [member=19219]clayop[/member] .
One more thing need to be considered add into analysis: If we apply % fee to an asset, hopefully quantity of micro payments in that asset will increase, and perhaps quantity of big payments will decrease. If we make an estimation for example apply *5 to the former and 1/3 to the latter(depends on the parameters change), how will the result change?

//For example, 30BTS is unacceptable for most of tipping UIAs, so there are very few tips in the historical data. If we set the fee to 1BTS, perhaps tipping would get much more active.

I will do it when I have time. But I think big transfers won't decrease much, because they are real demands. On the other hand, we can expect more micro-transactions, but I cannot guess the number because if some services using micro-payment (e.g. messaging / telecommunications) are really successful, they can increase the number dramatically. Assuming 10,000 messages are sent a day, the total fee goes up from 600k to 1.1mil BTS.
Yes, you got my idea definitely. That's why we tried to design a percentage based fee structure.
Charge a little more on real demands, charge less on small payments to attract more uses.
BTS account: abit
BTS committee member: abit
BTS witness: in.abit

Offline clayop

  • Hero Member
  • *****
  • Posts: 2033
    • View Profile
    • Bitshares Korea
  • BitShares: clayop
Re: poll for the percent based transfer fee
« Reply #13 on: January 21, 2016, 08:51:41 am »
Thanks for your hard work [member=19219]clayop[/member] .
One more thing need to be considered add into analysis: If we apply % fee to an asset, hopefully quantity of micro payments in that asset will increase, and perhaps quantity of big payments will decrease. If we make an estimation for example apply *5 to the former and 1/3 to the latter(depends on the parameters change), how will the result change?

//For example, 30BTS is unacceptable for most of tipping UIAs, so there are very few tips in the historical data. If we set the fee to 1BTS, perhaps tipping would get much more active.

I will do it when I have time. But I think big transfers won't decrease much, because they are real demands. On the other hand, we can expect more micro-transactions, but I cannot guess the number because if some services using micro-payment (e.g. messaging / telecommunications) are really successful, they can increase the number dramatically. Assuming 10,000 messages are sent a day, the total fee goes up from 600k to 1.1mil BTS.
Yes, you got my idea definitely. That's why we tried to design a percentage based fee structure.
Charge a little more on real demands, charge less on small payments to attract more uses.
+5%
With your development, we will be able to have decentralized message app with instant smartcoin tipping features ;)
Bitshares Korea - http://www.bitshares.kr
Vote for me and see Korean Bitshares community grows
delegate-clayop

jakub

  • Guest
Re: poll for the percent based transfer fee
« Reply #14 on: January 21, 2016, 11:13:31 am »
What fee are we talking about here? Trading fee or transfer fee?
Transfer fees only.