The Hilbert Guide
A cookie cutter approach to PVP

The Hilbert Files

An analysis of nearly a year of log files from Star Trek Online by Hilbert@mancom

 

The discussion about this data mining effort can be found on the STO PVP forums.

13
10
2012

Datamining

Shortly after the completion  of the initial version of the ACT plugin for STO I started keeping log files of all my PVP matches. Therefore I now have a huge collection of logs from random matches going back to December 2011. Prior to this I had only kept logs from special matches like tournaments.

I will attempt to analyse the data and show how the various abilities in STO have performed in the last year.

13
10
2012

The Development of Shield Healing 1: EPtS

We are going to take a look at how the shield healing abilities have evolved over time.

The obvious starting point is EPtS. Let's take a look at the distribution between EPtS 1/2/3. We can plot activations and the resulting heals. We see that the difference in magnitudes is comparatively small for EPtS and the charts don't differ much.

EPtS activations
Activations

EPtS heals
Heals

The distribution is sufficiently stable to take an average over the entire timeframe and compare it to the values from different premade matches.

EPtS heals in premade matches
EPtS heals in premade matches

A note regarding the premade matches: The 2012-05-18 game is the infamous 3h match. Apparently log files created in 2010 don't include shield healing, so the log files I have from KT league matches in 2010 will only become relevant once we get to hull healing (this means we will have some data points that predate the Borg set).

A quick calculation shows that the HPS (Healing Per Second) numbers are roughly what one would expect in a fight with lots of pressure. EPtS heals for something like 1200 per facing every 30s (since almost everyone carries two copies), which means that the theoretical HPS in a match is 1200*4*10 / 30 = 1600. If we keep in mind that actual HPS will be lower because of overhealing and the fact that not everybody (especially Sci/Sci or escorts) carries two copies, the recorded numbers for premade matches seem to be correct. We also see that in pug matches EPtS is not being used to its full potential.

We conclude that EPtS healing has been more or less constant since the beginning of 2011 and that EPtS1 is by far the most commonly used version of this skill.

14
10
2012

Science vs Engineering

As a reason for a recent change in the cooldown mechanics Borticus has stated:

We were attempting to encourage more use of Science-slotted healing skills, such as Hazard Emitters and Transfer Shield Strength, instead of every build focusing on the same set of Engineering heals in nearly every rotation (as currently exists).

Today, we shall take a look at whether this is justified by our data. We are going to compare the amount of science- and engineering-based healing for hull and shields in terms of both magnitude and activations. Since the changes are about bridge officer cooldowns, we will only look at bridge officer abilities and ignore things that are independent of a ship's bridge officer setup (e.g. RSF, BFI doff, Borg proc). Unfortunately we cannot see data for RSP since this ability is not part of STO's combat logs.

For shield healing, there are two engineering and two science abilities: EPtS/ES and ST/TSS. For hull healing there are AtSIF/ET and HE.

Now let's take a look at the distribution of the healing values between the abilities.

Science vs Engineering shield healing values
Shield heals

Science vs Engineering hull healing values
Hull heals

Almost 60% of the shield healing is done via science abilities (and TSS contributes significantly more than ST, but this is something we will investigate in more detail at a later point) and nearly 50% of the hull healing is based on Hazard Emitters which is the only science hull heal.

So going by healing values there does not seem to be a pressing need to shift healing more towards science abilities. But obviously AtSIF does not provide the same healing magnitude per activation as HE, so maybe there is a significant difference in how many copies of the different abilities are out there.

Let's take a look at activations. Since abilities like ES, TSS and HE do not provide all their healing in one tick, we need to normalise the recorded swings by a factor of 30 (for ES) or 15 (for TSS, HE). After doing this, we get the following charts that show the distribution of activations between the abilities.

Science vs Engineering shield healing clicks
Shield healing ability activations

Science vs Engineering hull  healing clicks
Hull healing ability activations

The picture is now a bit more in favour of engineering abilities. For shield heals where we have two of each type and comparable cooldowns (one 15/30 and one 30/45 ability) for both science and engineering, the shift is rather small and there is still no real bias towards engineering.

For hull healing, we see that science heals account for only one third of the activations. But since there are two engineering heals and only one science heal, this is to be expected. In fact, one would expect an even higher amount of engineering ability activations since these have significantly shorter cooldowns (15/15 and 15/30) than HE (30/45).

The data does not seem to support the idea that science-slotted heals need a boost by changing global cooldowns in their favour.

22
10
2012

The Development of Shield Healing 2: TSS

This time we are going to look at the development of TSS usage during the last year.

Once again we shall look at charts of activations and heals.

TSS Activations
Activations

TSS Heals
Heals

TSS2 is by far the most popular version of Transfer Shield Strength. It appears that the release of the Atrox carrier in May 2012 has led to a small increase in TSS3 use, but overall the differences are fairly small. (Looking ahead at partial data for 2012-10 indicates that the high value for TSS3 in 2012-09 does not remain at that level.)

As with EPtS, we can also compare the 2012 data with various premade matches.

TSS heals in premade matches
TSS heals in premade matches

Just like with EPtS we see values in premade matches that are more than twice as high as the average for the 2012 data set. The ratio between TSS1/2/3 depends a lot on the team composition in these matches. The SW/FS match had many science ships whereas the 2012-05-18 TSI/Pandas match primarily had cruisers in it.

25
10
2012

The Development of Shield Healing 3: ST

It is time for a look at Science Team. This skill used to be primarily employed as a means of clearing subnukes and scrambles, but lately it has seen widespread use as a high level healing skill.

Let's take a look at the usual charts.

ST Activations
Activations

ST Heals
Heals

While the use of ST2 has remained nearly constant, we see an increase in ST3 activity at the expense of ST1. The start of this trend in March appears to coincide with the release of the c-store Odyssey ships. The second bump in May coincides with the Atrox launch. Another effect that has probably increased the amount of ST3 is the fact that doffs allow the use of ST3 at its global cooldown while occupying only one LtC slot.

ST heals in premade matches
ST heals in premade matches

The difference between premade healing and pug healing is even bigger than it is for EPtS and TSS. This is most likely due to the fact that a majority of the ships in premade matches carry ST for SNB/SS-clearing whereas this is not the norm for players who don't usually engage in organised pvp.

It is interesting to note that the extremely sciship-heavy SW/FS match did not have any ST3 in it whereas the cruiser-heavy 2012-05-18 TSI/Pandas has a lot of ST3. This indicates a shift away from TSS3 as the major science shield heal for the LtC slot.

26
10
2012

The Development of Shield Healing 4: ES

It is time to look at Extend Shields. The use of this skill has considerably changed over time because of the various shifts in dominating ship classes and boff setups.

We'll start with the usual charts.

ES Activations
Activations

ES heals
Heals

We see that ES3 was the main flavour of extend at the end of 2011. This changed massively towards ES2 as the primary extend version and also ES1 saw increased use. This is most likely related to the increasing levels of non-extend shield resists from the MACO shield and BFI doff and also due to the fact that the doffs that allow the use of one copy of ET3 at its global cooldown have freed up a slot on cruisers that was previously occupied by ET2.

ES heals in premade matches
ES heals in premade matches

In premade matches we see a massive increase in ES-based healing in 2011(!). Interestingly there is also a huge difference between the two matches in 2011-11 that happened only three weeks apart.

On an absolute level, ES-healing has remained fairly stable in 2012 as the following chart shows. The spike in May is most likely due to the 3h TSI/Pandas match on 2012-05-18.

ES Heals in absolute numbers
Heals in absolute numbers

01
11
2012

The Development of Shield Healing 5: Borg Shield Proc and BFI Doff

In the 5th part of this series we will take a look at how the Borg shield proc (Multi-Regenerative Shield Array) and the BFI doff (Shield Distribution Officer) have shaped the game.

The magnitude of the borg shield proc was heavily reduced in S5 (to roughly 45% of its previous value per tick). This decrease is visible in monthly averages, but the weekly averages show the sudden decline even better. It is also visible that the reduction in average healing per second is much bigger than 55%. This is most likely due to the introduction of a hidden timer that prevents "chaining" of procs and maybe also because of people moving away from the 3-part Borg set to different sets.

Borg Shield Proc (months)
Heals (months)

Borg Shield Proc (weeks) Heals (weeks)

The data for premade matches (most of them in 2011) shows more of this development. Very high values for middle/late 2011 and low values for the match in 2012 due to the reduction in season 5. The first data point from March 2011 shows a very low value for the match. This is because at that time not many people were using the borg set. The magnitude of the proc (healing per tick) back then was as high as it was during the entire year 2011. But different deflectors and engines (the pre-nerf Efficient Engines in particular) were still in use at that time and not completely overshadowed by STF gear.

Borg Shield Proc heals in premade matches
Borg Shield Proc heals in premade matches

Since the Borg Shield Proc was removed in early 2012, this could have decreased the amount of shield healing in the game. But the massive increase in shield resist from the proc (now 50%), the fact that it now only triggers when needed (at least one shield facing needs to be at low health) and the introduction of a new shield heal in the form of the BFI doff negated that effect.

Let's take a look at the development of BFI doff healing.

BFI doff heals
Heals (months)

The monthly data shows an increase in BFI healing and a massive spike in March 2012. Once again we can get a better look at this effect in the weekly data. For a better look at the monthly data, we will also have a chart where we cut off the 2012-03 column.

BFI Doff heals (weeks)
Heals (weeks)

BFI Doff heals (months)
Heals (months, cut off in March)

The massive spike in March lasted only for a few days. It was created by a bug in the newly released Jem'Hadar shield that permanently triggered the BFI doff. Not many players abused this, but the few who did accumulated up to 100m healing per match and thus had a huge influence on the weekly and monthly averages.

Overall the BFI doff proc is hovering at slightly more than 1000 HPS during the last months. This is almost twice as high as the 600 HPS from the borg proc in late 2011. (Values for premade matches are obviously higher. The 2012-05-18 TSI/Pandas match has 2144 HPS from BFI.) To make things worse, the BFI doff is triggered manually, so in contrast to the 2011 Borg proc it is never wasted in situations where the heal is not really needed to survive.