trackobot.com
trackobot.com copied to clipboard
stevschmid
Stats for on the coin/play?
New user to track-o-bot, and can't figure out how to determine my winrate with a deck or over a time period with the coin vs without the coin. This is a stat I like to know, and wondering if it's been considered in the past.
I have little Ruby experience (mostly Python for web development work) but wouldn't mind taking an attempt at the feature if there's interest.
@jdharms if you query for the cards played in each match, it should be trivial to simply count the number of games where you play the coin or your opponent plays the coin (probably throw out any matches where no coin was played). AFAIK, there is no way of doing this from the UI and I doubt @stevschmid would be interested in adding that feature.
@theanine our definitions of trivial look like they're very different. That sounds like quite a bit of work, haha. It sounds like you're suggesting downloading the json data for a month or whatever and writing a script to tally the games, or manually going through the games in the ui (eeeeeek!).
It's fine if there's not interest in adding the feature, after all, it is something I can do myself with the API that's available. I just noticed that the API to post games to trackobot and the history API both have a field for "coin" for each game, but it doesn't appear to be displayed/filterable at all in the UI. This made me think there might be interest in displaying it somewhere.
So yes, different definitions of trivial, I suppose.
For what it's worth, other trackers seem to natively track this stat. So, it seems like something people want, or at least something a few developers thought people might want, and I thought I'd bring it up.
I know I'd prefer having access to the information even when I can't run Python scripts or when I'm on mobile.
Anyway, thanks for your help.
I'm curious what you consider to be trivial? I would think 34 LoC / 5-10min is certainly within trivial.
Also, just because other trackers have it doesn't mean it's a good feature to have. @stevschmid's philosophy is to keep ToB very minimal. And I think there are many more statistically-related features that would be much more useful than this measure that are deliberately not included in ToB (e.g. average time to play each deck type).
The only potentially useful thing about the feature you'd like would be to see which archetypes are more or less reliant on getting coin / going first. VS has an article on this (from their supply of many many more games than a single player has in their individual history) and that article isn't even very useful, because it basically shows exactly what you'd expect -- Miracle likes coin, aggro/zoo likes going first.
First, please don't assume everyone writes code as fast as you. Second, your solution doesn't provide the stats for each deck with the coin, or against each class. Maybe it's my math background poking through, but trivial means basically no work, not "anyone comfortable with Python can do it within an hour".
Tristan, are you officially connected to this project in any way? I understand that Steve's philosophy here is to keep things minimal, I just don't think that adding one filter option necessarily is in conflict with that.
I don't know if it's what you're trying to do, but you are thoroughly discouraging me from working on this feature, if it's even wanted, but also anything else for this project. And if this is the treatment I'll get when I open a ticket I will probably not be doing that again either.
Frankly, why are you being so antagonistic here? It doesn't look like you do this on every issue opened, so I'm wondering what it was about mine that caught your eye. Why do you presume to know "the only potentially useful thing" about a stat? Do you really not believe it's possible to improve your play with this information? What if someone found they had a 20% winrate difference without the coin? This almost certainly indicates poor usage of the coin, not just inherent weaknesses of the deck. This information is useful!
If you are someone who makes decisions about this projects development, please let me know so I can stop wasting your time. If you are not, please stop trying to convince me I don't need the feature I want. I'll gladly respect whatever direction Steve wants to go in with his project, but I would very much like him to consider it.
On Mon, Feb 13, 2017 at 3:28 AM Tristan Muntsinger [email protected] wrote:
I'm curious what you consider to be trivial? I would think 34 LoC / 5-10min is certainly within trivial.
Also, just because other trackers have it doesn't mean it's a good feature to have. @stevschmid https://github.com/stevschmid's philosophy is to keep ToB very minimal. And I think there are many more statistically-related features that would be much more useful than this measure that are deliberately not included in ToB (e.g. average time to play each deck type).
The only potentially useful thing about the feature you'd like would be to see which archetypes are more or less reliant on getting coin / going first. VS has an article on this (from their supply of many many more games than a single player has in their individual history) and that article isn't even very useful, because it basically shows exactly what you'd expect -- Miracle likes coin, aggro/zoo likes going first.
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/stevschmid/trackobot.com/issues/79#issuecomment-279322383, or mute the thread https://github.com/notifications/unsubscribe-auth/AA_mJvsv0kbc5agBXd-khEgETJ0fZn7Eks5rcBQegaJpZM4L-ZCM .
-- --Daniel
I'm a minor contributor to the project and @stevschmid doesn't get a lot of time to work on it, so I try to help where I can (on Reddit and here on GitHub) based on what I know. I also tried to help you by writing the code I thought trivial (and yes, I do consider "trivial" to be ~5 minutes of work; 0 minutes of work is a completed feature). I even wrote it in your preferred language of Python. I'll leave it to @stevschmid to respond to this issue, but as I said the interface is minimal by design as @stevschmid has stated quite a few times on here and on Reddit.
Perfect! Thanks for your help again. Appreciate the effort to package your sample code in a way readily accessible to me. I'd still much rather not have to run a script every time I want this information, and I think it could be useful to others as well.
I'm sorry for snapping at you--no one's getting paid here or owes me anything. I just felt like my feedback/suggestion was being discarded before it could even be considered, and felt frustrated.
On Mon, Feb 13, 2017 at 4:45 PM, Tristan Muntsinger < [email protected]> wrote:
I'm a minor contributor to the project and @stevschmid https://github.com/stevschmid doesn't get a lot of time to work on it, so I try to help where I can (on Reddit and here on GitHub) based on what I know. I also tried to help you by writing the code I thought trivial (and yes, I do consider "trivial" to be ~5 minutes of work; 0 minutes of work is a completed feature). I even wrote it in your preferred language of Python. I'll leave it to @stevschmid https://github.com/stevschmid to respond to this issue, but as I said the interface is minimal by design as @stevschmid https://github.com/stevschmid has stated quite a few times on here and on Reddit.
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/stevschmid/trackobot.com/issues/79#issuecomment-279532536, or mute the thread https://github.com/notifications/unsubscribe-auth/AA_mJkF_eExLaSXIHCPdNuVB1WU0R6-tks5rcM7kgaJpZM4L-ZCM .
This commit should be more useful for you. This should have everything you need. Also, this is another reason why ToB need not have the features baked in: with API access you can just play with the data as you see fit.
Also I'm sure there's a more pythonic way to write the code linked above, but I'm no Python expert. I'll accept any refactor PR.