The EXSTORM project is starting to deliver results. We've been running full-scale
in silico simulations of local energy market trading, and testing the viability of different trading heuristics. For example, we've put a good deal of work into the possibilities of posted-price trading, and the various heuristics for setting offer-to-sell prices and buy-limit prices.
We set out expecting to find that Derivative Follower (DF) heuristics, which work well in many kinds of market, would do a good job in a LEM. However, we somewhat painfully discovered that the fluctuations in energy abundance in a LEM have features that defeat ordinary DF pricers. We moved on to generalize from the principles of DF, developing a family of Generalized DF heuristics which appears to be much more robust. Here's a snapshot of a test with Generalized DF.
The blue lines show the abundance of energy in a LEM fluctuating (arbitrarily, for test purposes) between a large value, a small value and zero. The red dots show the sell-prices of five distributed-generation local vendors, and the green dots show five local buyers' ceiling prices (we limited the test run to ten participants, just to keep the snapshot clear: we usually run with 200 participants). You'll see how the participants' price expectations first moved to roughly align with each other, and then began to respond as an ensemble to the variation in energy abundance.
