Monday, September 30, 2013

CFC Interdiction Follow-Up

This was an interesting project.  Thankfully, some recent tools made the work go much smoother, and I'm really starting to sink my teeth into SAS JMP.  This piece was a lot of fun to write, since the overall theme from the graphs is "CFC failed to make sustained impact to the market", and TMDC is so widely regarded as only a propoganda hole.

Cutting Room Floor: Mining Report

This has been a topic I've wanted to cover, but haven't had the time to really cover it.  Also, the story is pretty brief.  
The story I wanted to tell was "ice miners, when harassed, will switch products", but the interdiction was not strong enough to force that to happen.  The story instead is a slow and steady decay in mineral prices.  It's interesting to see ice overtake minerals again, but I am not completely sure of the reasons.  When I tried to expand the story, I felt that the graphs were "show"ing more than I could really "tell".

Personally, I've been blaming the anemic market on the end-of-summer doldrums.  Unfortunately, I didn't know how how extensive the lull was.  
Graph Credit Ripard Teg at Jester's Trek
In my opinion, I think this is the most blunt graph to showcase on the state of EVE as a whole.  With no major conflict (CFC v TEST in Fountain drove the post-Odyssey spike) and a general malaise in accepting the status quo, my predicted rebound will probably fall far short of my personal expectations.

If I were to speculate in larger terms, I'd say that the slow slump in mineral prices is due to supply remaining constant while demand shrinks.  I could even further speculate that though the number of participants is declining, that there is some supply floor furnished by bots.  I admire CCP's fight against bots, but in a world where eradication is impractical, you are forced to take them into account at some level.  Unfortunately, I have no way to track bot numbers from any of my data feeds, so this effect is 100% speculative.  If demand picks up, I expect the price to flatten out and slowly rise, since full time mining is boring and not as lucrative as other endeavors, but unless player traffic drastically improves, I expect these swings to be very slow.

It's important to remember that we all exist in a circle of life between PVP and industry.  If PVP isn't destroying, industry has no one to sell to; if PVP outpaces industry, the price to PVP goes too high, and people become risk adverse.  My hopes are as we go into the winter, and through Rubicon, that moon incomes will recover to a level that CFC space starts to look enough to start prodding.  Also, I wonder if CFC leadership is intentionally spreading the forces thin just to test what the real end of their reach is.  

Graph Magic

I wanted to step up the quality of the graphs this time around.  Thanks to JMP, I was able to play with some best-fit tools.  

Using the default "best fit" method (or Smooth) isn't a great fit.  Though it is better than nothing, the fact that graph is so stark between periods, the smooth fit just isn't as useful.  Thankfully to a presentation at work, I looked up Kernel Smoother, which gives a bunch of knobs for how to best fit the data.  By changing the fit function to cosine and lowering the weight to better match the data, I'm able to put much better fit lines.  The best part is I can export the results from my best fit fiddling and move those to graph in any way/shape/form.  You'll see the numbers in the raw dumps.

This was exceptionally useful in my favorite graph of the piece, the Tags4Sec graph.  as a reference, this is kinda how it started:
Volume data is particularly spiky.  Being able to line up the wider trend of peaks and valleys I think gave a much more useful picture of the data rather than overlaying that noisy base data.  When I add the default smooth line, it's pretty clear that line tells the "macro" story, but isn't really useful for this project's analysis.  Also, the "repair rate" was an attempt to combine 3 volumes together in appropriate parts.  In this case, it was 1 part Trainer, 1.5 parts Recruiter, and 1.5 parts Transporter.  In retrospect, I could have been more creative with the combination, since I seriously doubt people were going all the way to -10 in ganking, so equal parts Recruiter/Transporter with no Trainer would have been a better picture.

What's Next?

Economic topics are running pretty lean lately.  There is the noise about PLEX prices skirting 600M and then dropping like a stone, thanks to an Amazon sale.  Also, I've been requested to analyze the "special edition" assets, to try and make a ruling on whether it's best to hold onto it as a collectors item, or dump in the initial release frenzy.  Beyond that, I'm out of topics until Rubicon shows up.

Also, though I am loving how much JMP is enabling me to get these sexy graphs out for people to enjoy, there's a licence problem: I have access to JMP at work, but players aren't going to buy keys.  I'd like to ween myself onto tools like R, but there's a lot of code between where I am today and being able to slice-n-dice as easily in a new set of tools.