Skip to content

Lane Analysis Tools

Detailed laning phase analysis tools for tracking CS, harass, tower proximity, rotations, and wave manipulation.

get_lane_summary

Get comprehensive laning phase summary with all tracked events.

This is THE tool for questions about "who won the lane", "laning phase summary", "how did the laning go", or general laning performance questions.

get_lane_summary(match_id=8461956309)

Returns: 

{
  "success": true,
  "match_id": 8461956309,
  "top_winner": "dire",
  "mid_winner": "radiant",
  "bot_winner": "radiant",
  "radiant_laning_score": 285.5,
  "dire_laning_score": 312.0,
  "hero_stats": [
    {
      "hero": "antimage",
      "team": "radiant",
      "lane": "bot",
      "role": "core",
      "last_hits_5min": 28,
      "last_hits_10min": 65,
      "denies_5min": 3,
      "denies_10min": 8,
      "gold_5min": 2100,
      "gold_10min": 4800,
      "level_5min": 5,
      "level_10min": 10,
      "lane_efficiency": 0.85,
      "damage_dealt_to_heroes": 450,
      "damage_received_from_heroes": 620,
      "time_under_own_tower": 45.5,
      "time_under_enemy_tower": 12.3,
      "last_hit_events": [...],
      "harass_events": [...],
      "neutral_aggro_events": [...],
      "tower_pressure_events": [...],
      "neutral_attacks": 50,
      "pull_attempts": 12,
      "tower_damage_taken": 450,
      "tower_hits_received": 3
    }
  ],
  "rotations": [...],
  "wave_nukes": [...],
  "neutral_aggro": [...],
  "tower_pressure": [...]
}

Key Fields:

Field Description
top_winner, mid_winner, bot_winner Lane winner: "radiant", "dire", or "even"
hero_stats Per-hero laning stats including CS, gold, harass
lane_efficiency Lane efficiency score from OpenDota (0.0-1.0, gold earned vs max possible). New in SDK 7.40.1
time_under_own_tower / time_under_enemy_tower Seconds spent under tower (lane equilibrium indicator)
last_hit_events Detailed list of every last hit/deny
harass_events List of hero-to-hero damage events
rotations Smoke breaks, TPs, twin gate usage
wave_nukes AoE ability usage on lane creeps
neutral_aggro All hero attacks on neutral creeps (pulls, farming)
tower_pressure All tower damage taken by heroes
neutral_attacks Count of neutral creep attacks per hero
pull_attempts Count of pull-like neutral aggro per hero
tower_damage_taken Total tower damage received per hero
tower_hits_received Number of tower hits per hero

Lane Efficiency

The lane_efficiency field (0.0-1.0) from OpenDota measures gold earned vs theoretical maximum:

  • >0.8: Excellent laning, free-farming
  • 0.6-0.8: Good laning, some pressure
  • 0.4-0.6: Contested lane
  • <0.4: Struggled in lane, likely lost

get_cs_at_minute

Snapshot of CS data for all heroes at a specific minute.

get_cs_at_minute(match_id=8461956309, minute=10)

Returns: 

{
  "antimage": {"last_hits": 65, "denies": 8, "gold": 4800, "level": 10},
  "medusa": {"last_hits": 72, "denies": 5, "gold": 5200, "level": 11}
}

Example Questions:

  • "How many last hits did the carry have at 5 minutes?"
  • "Compare CS between the two mids at 10 minutes"
  • "Who had the most denies at minute 5?"

Lane Analysis Service Methods

The LaneService provides detailed lane analysis through these methods:

get_lane_last_hits

Get all last hit and deny events during laning phase with positions.

from src.services.lane.lane_service import LaneService

lane_svc = LaneService()
last_hits = lane_svc.get_lane_last_hits(parsed_data, hero_filter="antimage")

Returns: 

[
  {
    "game_time": 35.2,
    "game_time_str": "0:35",
    "hero": "antimage",
    "target": "npc_dota_creep_badguys_melee",
    "is_deny": false,
    "position_x": -5200.0,
    "position_y": -4100.0,
    "lane": "bot"
  }
]

Use for:

  • "Where did the carry farm each wave?"
  • "How many denies did the offlaner get?"
  • "Show all last hits in the first 2 minutes"

get_lane_waves

Get wave-by-wave CS breakdown for a specific lane.

from src.services.lane.lane_service import LaneService

lane_svc = LaneService()
waves = lane_svc.get_lane_waves(
    parsed_data,
    lane="bot",
    team="dire",
    hero_filter="juggernaut"
)

Parameters:

Parameter Type Default Description
data ParsedReplayData required Parsed replay data
lane str "bot" Lane to analyze: "top", "mid", "bot"
team str "radiant" Team whose creeps to track: "radiant" or "dire"
hero_filter str None Filter CS to specific hero
end_time float 600.0 End of laning phase (seconds)

Returns: 

[
  {
    "wave_number": 1,
    "spawn_time": 0.0,
    "spawn_time_str": "0:00",
    "first_death_time": 42.5,
    "last_death_time": 58.2,
    "lane": "bot",
    "team": "dire",
    "melee_deaths": 2,
    "ranged_deaths": 1,
    "total_deaths": 3,
    "last_hits": [
      {
        "game_time": 42.5,
        "game_time_str": "0:42",
        "hero": "juggernaut",
        "target": "npc_dota_creep_badguys_melee",
        "is_deny": false,
        "lane": "bot",
        "wave_number": 1
      }
    ]
  },
  {
    "wave_number": 2,
    "spawn_time": 30.0,
    "spawn_time_str": "0:30",
    "first_death_time": 73.1,
    "last_death_time": 88.0,
    "lane": "bot",
    "team": "dire",
    "melee_deaths": 3,
    "ranged_deaths": 1,
    "total_deaths": 4,
    "last_hits": [...]
  }
]

Use for:

  • "How many CS did the carry get on wave 1?"
  • "Show wave-by-wave farming efficiency"
  • "Which waves did they miss CS on?"
  • "Compare first 3 waves between two carries"

Wave Timing

Waves spawn every 30 seconds (0:00, 0:30, 1:00, etc.) but don't die immediately:

  • Wave 1 (spawns 0:00) → creeps typically die 0:35 - 1:05
  • Wave 2 (spawns 0:30) → creeps typically die 1:05 - 1:35
  • Wave N (spawns (N-1)30) → creeps die spawn+35 to spawn+65*

The service uses non-overlapping 30-second death windows to assign each CS to the correct wave.

Wave Analysis Patterns

Common wave analysis use cases:

  • Perfect CS check: Wave has 3 melee + 1 ranged = 4 total
  • First wave priority: Getting all CS on wave 1 sets the tempo
  • Lane dominance: Consistent 3-4 CS per wave indicates control
  • Recovery detection: Low CS early, improving later shows adaptation

get_lane_waves_v2

Enhanced wave detection using entity_deaths + attacks correlation (requires python-manta 1.4.5.4+).

from src.services.lane.lane_service import LaneService

lane_svc = LaneService()
waves = lane_svc.get_lane_waves_v2(
    parsed_data,
    lane="bot",
    team="dire",
    hero_filter="juggernaut"
)

What's different from get_lane_waves:

Feature get_lane_waves get_lane_waves_v2
Death detection Combat log entity_deaths collector
Last hit attribution Combat log DEATH events Attack correlation
All creep deaths Hero last-hits only ALL creep deaths tracked
Contested CS Not detected Can be correlated

Returns: Same CreepWave structure as get_lane_waves.

Falls back to get_lane_waves if entity_deaths collector not available.

get_contested_cs

Detect contested CS - creeps attacked by 2+ heroes before death.

Uses entity_deaths + attacks correlation to find creeps where multiple heroes competed for the last hit.

contested = lane_svc.get_contested_cs(parsed_data, lane="bot", team="dire")

Returns: 

[
  {
    "game_time": 45.2,
    "game_time_str": "0:45",
    "entity_id": 1234,
    "creep_name": "npc_dota_creep_badguys_melee",
    "wave_number": 1,
    "hero_attackers": ["juggernaut", "axe"],
    "last_hitter": "juggernaut",
    "total_attacks": 8,
    "hero_attacks": 5
  }
]

Use for:

  • "How many creeps were contested in lane?"
  • "Who won the contested last hits?"
  • "Show trades where both heroes went for the same creep"

Requires python-manta 1.4.5.4+

This feature requires the entity_deaths and attacks collectors. Returns empty list if collectors not available.

get_lane_harass

Get hero-to-hero damage events during laning phase.

harass = lane_svc.get_lane_harass(parsed_data, hero_filter="lion")

Returns: 

[
  {
    "game_time": 42.5,
    "game_time_str": "0:42",
    "attacker": "lion",
    "target": "antimage",
    "damage": 65,
    "ability": "lion_impale",
    "lane": "bot"
  }
]

Use for:

  • "How much harass did the support deal?"
  • "Who took the most damage in lane?"
  • "Show me trading patterns in top lane"

get_tower_proximity_timeline

Track when heroes enter/leave tower range (detected via modifier_tower_aura_bonus).

tower_events = lane_svc.get_tower_proximity_timeline(parsed_data, hero_filter="antimage")

Returns: 

[
  {
    "game_time": 45.0,
    "game_time_str": "0:45",
    "hero": "antimage",
    "tower_team": "radiant",
    "event_type": "entered"
  },
  {
    "game_time": 52.0,
    "game_time_str": "0:52",
    "hero": "antimage",
    "tower_team": "radiant",
    "event_type": "left"
  }
]

Use for:

  • "How long was the carry under tower?"
  • "When did the lane push into enemy tower?"
  • "Track lane equilibrium via tower proximity"

Tower Aura Detection

Tower proximity is detected via the modifier_tower_aura_bonus buff which applies to heroes within 700 units of their own tower. This detects when heroes are near their allied tower.

get_wave_nukes

Detect when heroes use AoE abilities to damage multiple lane creeps.

nukes = lane_svc.get_wave_nukes(parsed_data, hero_filter="pugna")

Returns: 

[
  {
    "game_time": 28.5,
    "game_time_str": "0:28",
    "hero": "pugna",
    "ability": "pugna_nether_blast",
    "creeps_hit": 4,
    "total_damage": 280,
    "lane": "top"
  }
]

Use for:

  • "Did the support push the wave?"
  • "When did they nuke the creep wave?"
  • "Track intentional wave manipulation"

get_lane_rotations

Detect rotation events: smoke breaks, TP scrolls, twin gate usage.

rotations = lane_svc.get_lane_rotations(parsed_data, hero_filter="lion")

Returns: 

[
  {
    "game_time": 180.0,
    "game_time_str": "3:00",
    "hero": "lion",
    "rotation_type": "smoke_break",
    "from_position_x": -4500.0,
    "from_position_y": 800.0,
    "to_lane": "mid"
  },
  {
    "game_time": 320.0,
    "game_time_str": "5:20",
    "hero": "earthshaker",
    "rotation_type": "tp_scroll",
    "from_position_x": 2100.0,
    "from_position_y": -1500.0,
    "to_lane": null
  }
]

Rotation Types:

Type Description
smoke_break Smoke revealed (hero entered enemy vision/proximity)
tp_scroll Hero started teleporting
twin_gate Hero used twin gate portal

get_neutral_aggro

Get all hero attacks on neutral creeps during laning phase (0-10 minutes).

neutral_events = lane_svc.get_neutral_aggro(parsed_data, hero_filter="nevermore")

Returns: 

[
  {
    "game_time": 140.5,
    "game_time_str": "2:20",
    "hero": "nevermore",
    "target": "npc_dota_neutral_centaur_khan",
    "damage": 85,
    "camp_type": "large",
    "near_lane": "mid"
  }
]

Use for:

  • "Who was farming neutrals during laning?"
  • "Did the support pull the lane?"
  • "How much did SF farm neutrals mid?"

Camp Types:

Type Description
small Small camps (kobolds, hill trolls)
medium Medium camps (wolves, ogres, harpies)
large Large camps (centaurs, satyrs, hellbears)
ancient Ancient camps (black dragons, thunderhides)

Pull Detection

Events with near_lane set indicate the hero was near a lane when attacking neutrals - this often indicates pull attempts by supports.

get_tower_pressure

Get all tower damage taken by heroes during laning phase.

tower_events = lane_svc.get_tower_pressure(parsed_data, hero_filter="naga_siren")

Returns: 

[
  {
    "game_time": 85.0,
    "game_time_str": "1:25",
    "tower": "npc_dota_goodguys_tower1_bot",
    "hero": "naga_siren",
    "damage": 230,
    "tower_team": "radiant",
    "lane": "bot"
  }
]

Use for:

  • "Who took the most tower aggro?"
  • "Was the offlaner diving tower?"
  • "How much tower damage did the carry take?"

Tower Aggro Analysis

High tower damage in laning phase often indicates:

  • Offlaner playing aggressively (normal)
  • Carry being pushed under enemy tower (lane lost)
  • Dive attempts by supports or mid

Lane Analysis Example Queries

"Who won each lane?" 

summary = get_lane_summary(match_id=8461956309)
# Returns: top_winner, mid_winner, bot_winner

"How did the carry's lane go?" 

summary = get_lane_summary(match_id=8461956309)
carry = next(h for h in summary.hero_stats if h.role == "core" and h.team == "radiant")
# Check: last_hits_10min, damage_received, time_under_own_tower

"Did the support rotate mid?" 

summary = get_lane_summary(match_id=8461956309)
mid_rotations = [r for r in summary.rotations if r.to_lane == "mid"]

"Was the wave pushed into tower?" 

tower_events = lane_svc.get_tower_proximity_timeline(data, hero_filter="antimage")
# Track entered/left events to see lane equilibrium

"Did the support pull the lane?" 

summary = get_lane_summary(match_id=8461956309)
support = next(h for h in summary.hero_stats if h.role == "support" and h.lane == "bot")
# Check: pull_attempts, neutral_attacks

"Who was farming neutrals mid?" 

summary = get_lane_summary(match_id=8461956309)
mid_neutral_aggro = [na for na in summary.neutral_aggro if na.near_lane == "mid"]
# Group by hero to see who farmed most

"Who took the most tower damage?" 

summary = get_lane_summary(match_id=8461956309)
# Sort hero_stats by tower_damage_taken
most_tower_damage = max(summary.hero_stats, key=lambda h: h.tower_damage_taken)

"Show me the carry's first 3 waves" 

from src.services.lane.lane_service import LaneService

lane_svc = LaneService()
waves = lane_svc.get_lane_waves(data, lane="bot", team="dire", hero_filter="juggernaut")
for wave in waves[:3]:
    print(f"Wave {wave.wave_number}: {wave.total_deaths} CS")
    # Wave 1: 3 CS, Wave 2: 4 CS, Wave 3: 2 CS

"Compare wave-by-wave efficiency" 

waves = lane_svc.get_lane_waves(data, lane="bot", team="dire", hero_filter="juggernaut")
total_possible = len(waves) * 4  # 4 creeps per wave
total_cs = sum(w.total_deaths for w in waves)
efficiency = total_cs / total_possible * 100
# "Juggernaut got 85% of possible lane CS"

Data Limitations

Currently NOT trackable from replay data:

Data Status Why
Tower damage on creeps Missing Combat log doesn't include tower-to-creep damage
Creep-on-creep damage Missing Only hero interactions logged
Creep entity positions Missing Only hero positions tracked in entities
Creep entity IDs Missing Combat log uses generic names like npc_dota_creep_goodguys_melee
Lane equilibrium position Derived Approximated via tower proximity
Stack count detection Partial Can count clears but not stacks created

Lane Equilibrium Analysis

While exact creep positions aren't available, you can analyze lane equilibrium through:

  • Tower proximity timeline: Time spent under own vs enemy tower
  • Wave nuke detection: When heroes push the wave
  • Hero position at CS: Where last hits occurred
  • Harass patterns: Trading forces lane to push/pull