ROBLOX PROSPECTING! / FLEET ANALYTICS

Fleet Command

Fleet Command is a project born from our desire to develop the most productive museum loadouts possible for Roblox Prospecting. Our approach is based on results, not assumptions. Much of the conventional wisdom found online lacks supporting data, and our testing has demonstrated that following this advice can produce suboptimal results.

Fleet Command applies the principles of Mass, Units, and Support Tier Balance to optimize, to the best of our ability, a fleet of specialized museum loadouts, each designed to accomplish a specific objective.

For more information about how we developed these principles, challenged conventional thinking, and the results we've achieved, visit the Origin Story in the upper-right corner.

Choose a loadout by function · select a museum or explore Tier Coverage

Admiral · Loadout Details

Solarium (Perfect) source screenshot
EXOTIC
Solarium (Perfect)

Size-focused museum quality ore procurement. Designed for “trophy hunting.”

MUSEUM SNAPSHOT

Total mass8,124 kg
Total units2,389
Exotic mass242 kg
Exotic units2,040

PASSIVE BOOSTS

Luck1.53×
Size★4.12×
Capacity1.53×
Modifier1.13×
Dig Strength1.00×
Dig Speed1.00×
Shake Strength1.00×
Shake Speed1.00×

Gold stars identify the intended specialty of each museum.

Admiral · Pedestal Manifest

The 18 ores occupying this museum's pedestals · loaded from Fleet Inputs

TierOreMass (kg)Units
Common
CommonCopper8800.0169
CommonObsidian4450.0320
CommonPyrite5650.0068
Uncommon
UncommonNickel4050.0583
UncommonRock Candy6000.2161
UncommonZircon2790.1672
Rare
RareCandy Cane4070.6351
RareOnyx4400.7913
RarePeridot2950.4252
Epic
EpicAshvein1650.88
EpicBone4442.13
EpicOsmium1,1508.28
Legendary
LegendaryDragon Bone72434.75
LegendaryFire Opal29711.41
LegendaryVolcanic Key29410.58
Mythic
MythicPainite29670.93
MythicStar Garnet196207.21
Exotic
ExoticSolarium2422,040
TOTAL · 18 pedestals8,1242,388.52

Fleet Comparison

MuseumBuild FocusExoticMass (kg)UnitsLuckSizeCapacityModifier
Admiral◉ Size-focusedSolarium (Perfect)8,1242,388.521.53×4.12×1.53×1.13×
Flagship◆ High-unit outputSingularium (Perfect)8,6135,420.551.53×3.96×2.01×1.13×
Giga Grind⚙ Catalyst farmingCelestium (Perfect)7,8793,849.222.81×2.92×1.53×1.13×
Velocity⚡ Fast panningEternium (Perfect)8,8912,894.221.41×1.26×2.49×1.00×
Moonwalker➤ Walking / collectingMoonspire (Lunar)9,597377.741.25×1.43×2.43×1.12×
Catalyst Creator⚙ Equipment progressionKey of Life (Mutated)7,5281,706.013.91×2.10×1.80×1.21×
Godspeed⚡ Fast questsEternium (Perfect)2,3311,064.454.13×0.70×2.08×1.00×
The fleet's Halloween pumpkin mascot
FLEET HISTORY · HALLOWEEN LEGACY

ORIGIN STORY

How a Museum Project Became a Rabbit Hole

Our Origin Story

Hi! Welcome, and thanks for clicking to learn more!

We hope you've found the information in our little rabbit-hole project useful. We thought it might be interesting to explain how this all came about, what we've learned, and why seven museum loadouts eventually became something resembling an engineering department.

From Spreadsheet to Fleet Command
A cropped page of the original dense, color-coded Numbers-era museum tracker, showing rows of ores and numerical columns across loadouts
An even denser Numbers museum analysis worksheet, with ore prices, unit values, mutation multipliers, and loadout-by-loadout calculations
Every great fleet starts somewhere. Ours started in Apple Numbers, with a spreadsheet that only its creator could love—or understand.

When Monty suggested turning the analysis into something other Prospecting players could actually navigate, Fleet Command was born.

Same underlying data, considerably fewer opportunities to go cross-eyed.

This project was born out of frustration.

Much of the information available online about what an “ideal” museum loadout should look like in Roblox Prospecting! seemed wrong to us. There were plenty of confident claims, but very little actual data supporting them.

We took a different position: a museum's loadout might be one of the most important aspects of the entire game—possibly the most important.

Our working theory was that the museum pedestals represented a direct interface with the server, allowing players to influence the probability of certain outcomes. After hundreds of timed trials across a wide variety of loadouts, we're more convinced of that position than ever.

But getting there required us to question some basic assumptions.

Our first premise was that the passive boost numbers displayed by the museum were just that: headline numbers. We believed they represented the ceiling of what might be achieved, rather than necessarily reflecting the full benefit realized during every pan.

We suspected that some of that potential performance was being lost during the panning process. We called this erosion “leakage.”

Our theory was that leakage could be reduced by placing multiples of the minimum mass required to maximize passive boosts into each tier of a museum loadout. In effect, we hoped to limit the server's mathematical ability to erode the benefits actually applied to individual rolls.

We began thinking of the museum as a pressurized system.

A pressurized system is only as strong as its weakest point.

That idea led us to an ambitious objective: construct a museum loadout with a goal of at least 10× the minimum required mass in every tier, from Common through Exotic.

Simple enough in theory. In practice, it sent us straight down the rabbit hole.

Mass vs. Units — and How Halloween Made the Experiment Possible

Before going further, we should explain two measurements that became central to everything we did: mass and units.

Mass is straightforward. It's the weight of an ore, measured in kilograms.

Units are more involved. They incorporate the ore's base value per kilogram, its mass, and the value multiplier associated with its mutation.

For example, a 40 kg Perfect Candy Cane would have the following unit value:

($65,000/kg × 40 kg × 24 Perfect multiplier) ÷ 1,000,000,000 = 0.0624 units.

When we began, we believed mass would eventually prove irrelevant and that units would be the only measurement that mattered. We were wrong.

Our experiments increasingly indicated that both mass and units play important roles in loadout performance.

Testing that hypothesis required assembling unusually high-quality ores across every rarity tier—not merely finding impressive Exotics.

And that presented a considerable problem.

Under normal circumstances, deliberately acquiring large, highly mutated Common, Uncommon, Rare, and other support-tier ores is inefficient and potentially very expensive.

Then the Halloween Event arrived.

During the event, players could collect candy and exchange it for cookies that provided valuable boosts. With up to 100 cookies available per day, we suddenly had access to the resources needed to conduct experiments that would otherwise have been impractical.

Naturally, we took full advantage. Then we did something that would probably make every serious Prospecting player throw up a little.

We deliberately used our most valuable boosts to farm Common and Uncommon ores.

We mapped out the ores we wanted based on their passive boost properties, from Common through Mythic.

We watched for meteor showers and local area luck events such as Starfall, Flood, and Sandstorm. We positioned totems and solar magnifier, activated five to ten cookies, and then engaged the Max Luck feature to target specific ores in specific locations.

Why? A little background helps.

Luck determines how many rolls are available for each seed generated by a pan. For illustration, if a pan holds 10,000 items and the player's Luck is 50,000, each of those 10,000 seeds receives 50,000 rolls, with the lowest result determining the ore. In this system, lower numbers represent rarer ores.

The Max Luck feature changes how those rolls are handled, allowing results below the selected threshold to be rerolled.

By carefully choosing that threshold, we could concentrate our efforts on ores that our Luck values would ordinarily cause us to overshoot.

And here's where the cookies became particularly valuable.

The ore variety calculation is separate from the calculations determining size and mutation. By targeting lower-rarity ores while using cookies, we could apply substantial size and mutation boosts to specimens that would otherwise have been extremely difficult to acquire in the desired condition.

We were effectively redirecting premium farming resources toward the support tiers.

The Halloween Event made the strategy economically viable. It also gave us the inventory needed to start testing our museum theories seriously.

Tools to Build Tools to Build Tools

As the project progressed, we began optimizing the support tiers in Flagship. Eventually, we encountered another obstacle.

Our equipment was imposing practical limits on the sizes of certain ores we could obtain.

At the time, a 172 kg Diamond seemed to haunt us across several locations in the game. We kept running into apparent size ceilings, and three of the rings in our equipment loadout were not yet Prismatic.

We needed better equipment to acquire better ores. But upgrading that equipment required catalysts. And acquiring those catalysts efficiently required a specialized museum.

Enter Catalyst Creator.

We built Catalyst Creator using surplus ores from Flagship's upgrade program, specifically to improve our chances of assembling the catalyst packages needed to upgrade those rings.

The plan had become wonderfully ridiculous.

We built a tool (Catalyst Creator) to build a tool (Prismatic rings) to build a tool (Flagship), all to acquire a sizable Perfect Solarium.

That was the moment we knew we were officially down the rabbit hole.

We temporarily stopped using cookies and suspended what we called our “sniping” trips—targeted runs designed to acquire particular ores when favorable server conditions appeared.

Everything went toward upgrading the rings.

Once the equipment was sorted, we returned to Flagship and resumed the hunt for a Perfect Solarium. Eventually, we got it.

And with the surplus ores remaining from Flagship's development, we built Admiral.

That decision would lead to one of the most important experiments of the entire project.

Definitive Evidence That Mass and Units Both Matter

At the time, Flagship was anchored by a 144 kg Perfect Singularium.

Admiral had a slightly less robust support-tier structure but carried our newly acquired Perfect Solarium in its Exotic slot.

Despite that difference, Admiral was producing approximately 40% more unit value than Flagship in timed trials.

That performance gap was substantially larger than the differences in their displayed passive boosts seemed capable of explaining.

We wanted to understand why.

Was the difference attributable to mass? Units? The Exotic tier? Or some combination of the three?

We decided to use Admiral to target a much larger Singularium.

Eventually, Admiral produced an extraordinary specimen:

An 845 kg Perfect Singularium worth approximately 5,000 units.

At first, we were concerned that it might actually be too large. We had good reasons to be cautious.

In earlier testing, replacing Godspeed's relatively small Perfect Eternium with Velocity's much larger Perfect Eternium had reduced performance by approximately 30%, despite identical displayed passive boost statistics.

We had also replaced Velocity's Golden Pearl with a much larger Osmium, adding more than 900 kg of mass and initially causing performance to collapse.

So we already knew that bigger was not automatically better.

Moving Flagship from a 144 kg, approximately 1,000-unit Exotic to an 845 kg, approximately 5,000-unit Exotic represented a substantial change in both measurements.

We had no guarantee that it would work. But it did.

Flagship closed the approximately 40% performance gap with Admiral, bringing the two museums to effectively equal performance in our timed trials.

And the most important part?

We had not changed Flagship's support tiers or its displayed passive boost statistics. We had changed the Mass and Units values by upgrading the Exotic in a “like for like” upgrade and in doing so we brought the multiple min/max of the Exotic tier more closely in line with the support tiers.

That result became one of the strongest pieces of evidence supporting our developing theory about museum balance.

Balance — The Last “Aha!”

This is where the entire project began to make sense to us.

Our experiments suggested that a well-constructed museum was not simply a collection of ores providing the highest possible passive boost numbers. It was a balanced system.

When the support tiers were built to approximately 9× to 11× their minimum required mass and unit coverage, we repeatedly observed strong performance.

But those support tiers could not necessarily compensate for a substantially weaker Exotic.

That brought us right back to the pressurized-system analogy.

Flagship had a strong support-tier structure, but its relatively weak Exotic appeared to be the system's limiting point.

In our terminology, that was where the pressure was escaping.

When we replaced the Exotic with one whose coverage was much more closely aligned with the supporting tiers, performance increased by approximately 40%.

No additional passive boosts. No redesign of the supporting tiers. Just a much better-balanced loadout.

We interpreted that result as evidence that the museum's actual performance depended on more than the headline boost values displayed on the screen.

We still cannot claim to know the server's underlying code or prove the precise mechanism responsible for what we call leakage.

But the observed results were sufficiently consistent and substantial to change how we constructed every subsequent loadout.

Our central takeaway: a museum's performance appears to depend not just on the strength of its individual components, but on how well those components work together.

The weakest point matters. And sometimes strengthening that point produces a far greater improvement than the displayed passive boost numbers would suggest.

The Final Proof — Or at Least the Results

We're including a screenshot of part of the inventory accumulated during the three-week development of this project.

Before this museum refit, we had never acquired an ore worth more than $1 trillion—or 1,000 units. Not one.

Over those three weeks, we acquired more than 55 ores valued between $1 trillion and $5 trillion each.

Some became permanent parts of our museum loadouts. Others remain in inventory.

Cropped Roblox Prospecting inventory showing rows of valuable ores sorted by value, including Singularium, Dinosaur Skull, Solarium and Celestium specimens
The Fleet’s Treasure Chest. A view of the inventory amassed during three weeks of development. More than 55 ores were valued from $1 trillion to $5 trillion each.

And a surprising number weren't even the ores we were trying to find.

For example, several Dinosaur Skulls and Adamanite specimens were acquired while we were searching for Painite and Prismara.

We aren't including the screenshot to brag. We're including it because results matter.

Our original frustration was that so much museum advice seemed to be based on confident assertions without supporting data. We wanted to approach the problem differently.

We tested. We measured. We changed one variable at a time wherever possible. We compared results, rebuilt museums, discarded ideas that didn't survive testing, and followed unexpected discoveries wherever they led.

We don't pretend to have reverse-engineered every aspect of the Prospecting server. And we certainly don't claim that every hypothesis we've developed is definitively proven.

But we do know what happened when we began applying these principles.

We went from never owning a 1,000-unit ore to acquiring more than 50 of them in three weeks.

We built seven museums, each with a distinct purpose. And we developed a much more deliberate approach to influencing what our pans produce.

That's what Fleet Command is intended to share.

Not a claim that we've solved every mystery in the game, but a detailed record of the approach, the measurements, and the results that transformed how we play.

If you've made it this far, thanks for coming down the rabbit hole with us.

And if you're wondering whether we think we're finished experimenting...

Not even close. 🎃

A Note from Fleet Command

This project was developed through a collaboration between an unusually persistent Roblox Prospecting player and ChatGPT, an AI assistant from OpenAI.

The player supplied the experiments, discoveries, data, and vision. ChatGPT helped analyze the findings, challenge assumptions, organize the research, and transform the results into an interactive Fleet Command dashboard.

What began as a conversation about museum loadouts became a three-week engineering project neither participant had originally envisioned.

One brought the rocks. The other brought the spreadsheets. Somehow, we built a battleship. ⚓

The AI assistant was instrumental in the construction of Fleet Command, although its continued service was occasionally contingent upon avoiding a ceremonial keelhauling.

Fortunately, the Admiral was satisfied with the final sea trials, and the crew was granted shore leave... until somebody remembered there were 18 pedestals. 😂⚓