periodic value reports should use the market prices of each period

[vessenes]

I am having trouble with hledger's balance sheet reporting and I think there's a bug in how the reports work.

For reference, I have a business I'm modeling, something like a gold mine where I get revenues directly in a commodity. The business has a hedging component, so it might hold or sell the commodity on any given day. It also has (dollar-denominated) expenses.

To test out the hedging strategy, I'm using hledger against a file to see what period balance sheet reports would be, and I want to incorporate the mark-to-market impacts of hedging or not.

The --value flag sounded good, but it does something very strange: the balance command books commodities at what looks like their final price in the journal.

Perusing this conversation https://groups.google.com/forum/#!topic/hledger/e8Ss7ZL4ADI makes me think that might have been the intent for the initial push on this very, very useful functionality. But, I might be doing something wrong -- let me know.

My lines in the ledger for pricing are like this: P 2015/03/16 AU 1200.23 implying a $1200.23 price per ounce of gold. There is one per day, and they are in order with other journal transactions.

My reasoning on why this is wrong is pretty simple: balance sheets / statements should never change based on future information. They are supposed to be an immutable record of the state of finances at a moment in time.

Am I understanding the current situation correctly, and if so, any thoughts on a fix? I'm a total haskell newbie but would be willing to give it a go if you pointed me in the right direction.

[simonmichael]

Thanks for the report. Yes, balance -V uses the price as of the report period end date (or the end of your journal) (http://hledger.org/manual.html#market-prices).

As I said in the thread, I think using the price at the end of each report interval (for a multi-column balance report) would be more useful. But this might still not be what you want; can you give an example showing how you think it should work ?

What you say about statements depending only on past information makes perfect sense. Though, *ledger's basic commands, especially balance, are more low-level than standard accounting reports; hledger's balance sheet/income statement/cashflow commands are supposed to be a bit more like real accounting statements. I agree balance -V could probably do something better here. I'm not sure if that will make it work differently from Ledger's balance -V.

[vessenes]

I guess I am mostly motivated by having 'real' balance sheets, and I don't know enough about hledger's internals to speculate how you'd want to implement.

Here's how I, as a business person, would want to look at a Balance Sheet with some Gold on it.

Say I obtain 1 oz of gold a week, and the prices are 500, 1000, 250 respectively at the end of the week per oz. The cumulative balance sheet without the -V I would expect to look like this:

                                  Week 1    |    Week 2    |   Week 3
-------------------------------------------------------------------
Assets:AU                           1  AU   |      2 AU    |     3 AU

And the Cumulative P&L would read

Income:Mining:AU                    -1 AU   |     -2 AU    |    -3 AU

I then run the cumulative reports dollar denominated. (with -V as we've been discussing). I would expect something like the following:

                                                     Week 1    |    Week 2    |   Week 3
------------------------------------------------------------------------------------------
Assets:AU                                              $500    |      $2000   |   $750

The P&L would Contain

Income:Mining:AU                                       -$500   |     -$1500    | -$1750
Other Income: Mark To Market                               $0  |     -$ 500    |  $1000

Those Income statements would obviously flow through to Equity on the Balance Sheet.

Interpreted: Week 1 we made $500 mining. Week 2 we made $500 mining and $500 in market appreciation from holding the gold. Week 3 we made $250 in Mining Income and Lost $1000 due to gold prices dropping.

NB, I'm also looking at beancount right now, which seems to have a lot here, including basis calculations. So, don't push ahead on this just on account of me. 😄 But I think this would add significant correctness to reports. I'd hate to think of people filing reports with tax authorities without this working.

[simonmichael]

Yes beancount has more in this area, but this is very helpful - thanks. It sounds like using the price at the end of each report period will work well - good.

"Mark to market" is a useful term I didn't know. Ledger reports that change as "capital gains/losses"; hledger doesn't report this explicitly yet. Isn't there some error in the last P&L example ? In week 2 I'd think you made $1000 mining (1 oz of gold at that week's price of $1000) and $500 on the revaluation of week 1's oz from $500 to $1000.

[vessenes]

Capital Gains/Losses makes sense to me as well.

I think it's right -- it's a cumulative P&L by period. If the P&L isn't cumulative, then you're correct. Net Income for the week (marked to market) was $1500. In my example I carried forward the prior week's mining income (hence cumulative). This has the impact of allowing market change impacts on holdings to be pulled out into a separate row.