The assertion is, as we onboard, this will put downward pressure on HIVE.  It's somewhat intuitive, but this assertion should be supported by evidence.

Usually, the only response I get is, "Why wouldn't it?"

This attitude contributes to the lack of marketing.  Why market something that will only result in downward pressure on the asset?

I would like to investigate the claim itself.  If my method is wrong, I would like to know where my faulty evidence is.

We should also keep in mind that any model we use will still be wrong because we're trying to model a network effect (positive or negative).

Strictly speaking, a model should be able to mitigate how wrong it is by error bars.  It doesn't have to nail the prediction.  Just model a possible outcome and see how far off it was from reality.

---

Let's start with definitions.  What do we mean by "dump"?  Do we mean transfers to an exchange?  This is reasonable.  If a user transfers assets to an exchange, the next likely step is to sell.  Let's ignore the rest of the implications around order book depth.  Just because someone sells doesn't mean the price will go down.  If there is simultaneous interest in buying as selling, the price doesn't drop.  But we're ignoring that, for the moment.

All we want to know is when an author is paid, are they likely to then transfer their liquid rewards to an exchange?

How do we determine this?

Well, we can look at the virtual operation known as `author_reward` which is represented by HiveSQL as `VOAuthorRewards`.

```sql
SELECT COUNT(*) FROM dbo.VOAuthorRewards r
  WHERE r.timestamp BETWEEN '2021-03-01' AND '2021-03-08'
=> 26665
```

In 26665 reward operations, we can also determine how many authors there were:

```sql
SELECT COUNT(DISTINCT(r.author))
  FROM dbo.VOAuthorRewards r
  WHERE r.timestamp BETWEEN '2021-03-01' AND '2021-03-08'
=> 4785
```

For that period, we know that the HIVE payout was zero.  So we can look at the HBD payout instead:

```sql
SELECT SUM(r.hbd_payout) FROM dbo.VOAuthorRewards r
  WHERE r.timestamp BETWEEN '2021-03-01' AND '2021-03-08'
=> 26416.2870
```

So we know that `26416.287 HBD` was paid in liquid rewards to 4785 authors.

This is what the price did in that timeframe:

<center><img src="https://files.peakd.com/file/peakd-hive/inertia/23tSzWNHEqiPAueGNbxtej3hiqaeDavXN9NgUkVNPJn6YLiByxCP79PDqYQnArn2QZP3A.png" /></center>

How many of those authors transferred HBD to Bittrex or Blocktrades?

```sql
SELECT COUNT(DISTINCT(t.[from])) FROM dbo.TxTransfers t
  WHERE t.timestamp BETWEEN '2021-03-01' AND '2021-03-08'
  AND t.[to] IN('bittrex', 'blocktrades')
  AND t.amount_symbol = 'HBD'
  AND t.[from] IN (
      SELECT r.author
      FROM dbo.VOAuthorRewards r
        WHERE r.timestamp BETWEEN '2021-03-01' AND '2021-03-08'
  )
=> 115
```

And how much HBD are we talking about?

```sql
SELECT SUM(t.amount) FROM dbo.TxTransfers t
  WHERE t.timestamp BETWEEN '2021-03-01' AND '2021-03-08'
  AND t.[to] IN('bittrex', 'blocktrades')
  AND t.amount_symbol = 'HBD'
  AND t.[from] IN (
      SELECT r.author
      FROM dbo.VOAuthorRewards r
        WHERE r.timestamp BETWEEN '2021-03-01' AND '2021-03-08'
  )
=> 3460.5970
```

---

Bottom line, `26416.287 HBD` was paid but only `3460.597 HBD` was sent to Bittrex or Blocktrades.

Remember, we're only asking if authors are generally likely to dump (e.g., we're *not* investigating the peg, here).  I would think, if authors dump, they would most likely dump HBD.  So the line of reasoning has become:

"More users means more dumps."

"Are you sure?"

"Why wouldn't it?"

"Because only ~ 3% of authors 'dumped' ~ 14% of the HBD earned in this one week I looked at."